doc/html/NEWS.2.html - R - Git at Google

 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><title>R: R News</title>
 <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
 <meta name="viewport" content="width=device-width, initial-scale=1.0, user-scalable=yes" />
 <link rel="stylesheet" type="text/css" href="R.css" />
 </head><body><div class="container">

 <table width="100%" summary="page for NEWS"><tr><td>NEWS</td><td style="text-align: right;">R Documentation</td></tr></table>

 <h2> R News</h2>

 <h3>LATER NEWS</h3>

 <p>News for <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 3.0.0 and later can be found in file &lsquo;<span class="file">NEWS.Rd</span>&rsquo; in the
 <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> sources and files &lsquo;<span class="file">NEWS</span>&rsquo; and &lsquo;<span class="file">doc/html/NEWS.html</span>&rsquo;
 in an <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> build.
 </p>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.15.3</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>lgamma(x)</code> for very small <code>x</code> (in the
 denormalized range) is no longer <code>Inf</code> with a warning.
 </p>
 </li>
 <li> <p><code>image()</code> now sorts an unsorted <code>breaks</code> vector,
 with a warning.
 </p>
 </li>
 <li><p> The internal methods for <code>tar()</code> and <code>untar()</code> do
 a slightly more general job for &lsquo;ustar&rsquo;-style handling of
 paths of more than 100 bytes.
 </p>
 </li>
 <li><p> Packages <span class="pkg">compiler</span> and <span class="pkg">parallel</span> have been added
 to the reference index (&lsquo;<span class="file">refman.pdf</span>&rsquo;).
 </p>
 </li>
 <li> <p><code>untar(tar = "internal")</code> has some support for
 <code>pax</code> headers as produced by e.g. <code>gnutar --posix</code>
 (which seems prevalent on OpenSUSE 12.2) or
 <code>bsdtar --format pax</code>, including long path and link names.
 </p>
 </li>
 <li> <p><code>sQuote()</code> and <code>dQuote()</code> now handle 0-length
 inputs.  (Suggestion of Ben Bolker.)
 </p>
 </li>
 <li> <p><code>summaryRprof()</code> returns zero-row data frames rather
 than throw an error if no events are recorded, for consistency.
 </p>
 </li>
 <li><p> The included version of PCRE has been updated to 8.32.
 </p>
 </li>
 <li><p> The <span class="pkg">tcltk</span> namespace can now be re-loaded after
 unloading.
 </p>
 <p>The Tcl/Tk event loop is inhibited in a forked child from package
 <span class="pkg">parallel</span> (as in e.g. <code>mclapply()</code>).
 </p>
 </li>
 <li> <p><code>parallel::makeCluster()</code> recognizes the value
 <span class="samp">random</span> for the environment variable <span class="env">R_PARALLEL_PORT</span>:
 this chooses a random value for the port and reduces the chance of
 conflicts when multiple users start a cluster at the same time.
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li><p> The default for <span class="env">TAR</span> on Windows for <code>R CMD
 	build</code> has been changed to be <span class="samp">internal</span> if no
 <code>tar</code> command is on the path.
 </p>
 <p>This enables most packages to be built &lsquo;out of the box&rsquo;
 without <code>Rtools</code>: the main exceptions are those which need
 to be installed to re-build vignettes and need <code>Rtools</code>
 for installation (usually because they contain compiled code).
 </p>
 </li></ul>


 <h4>C-LEVEL FACILITIES</h4>


 <ul>
 <li><p> On a 64-bit Windows platform with enough RAM, <code>R_alloc</code>
 can now allocate up to just under 32GB like other 64-bit
 platforms.
 </p>
 </li></ul>


 <h4>DEPRECATED AND DEFUNCT</h4>


 <ul>
 <li><p> Use of <code>col2rgb(0)</code> is deprecated (see the help page
 for its limitations).
 </p>
 </li>
 <li><p> The deprecated <code>intensities</code> component returned by
 <code>hist()</code> is no longer recognized by the <code>plot()</code> method
 and will be removed in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 3.0.0.
 </p>
 </li>
 <li> <p><code>real()</code>, <code>as.real()</code> and <code>is.real()</code> are now
 formally deprecated and give a warning.
 </p>
 </li>
 <li><p> This is formal notice that the non-API EISPACK entry points
 in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> will be removed shortly.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> The <code>configure</code> tests for Objective C and Objective
 C++ now work on Mac OS 10.8 with Xcode 4.5.2 (<a href="https://bugs.R-project.org/show_bug.cgi?id=15107">PR#15107</a>).
 </p>
 </li>
 <li><p> The cairo-based versions of <code>X11()</code> now work with
 current versions of cairographics (e.g., 1.12.10).  (<a href="https://bugs.R-project.org/show_bug.cgi?id=15168">PR#15168</a>)
 </p>
 <p>A workaround for earlier versions of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> is to use
 <code>X11.options(type = "nbcairo")</code>.
 </p>
 </li>
 <li><p> Configuration and <code>R CMD javareconf</code> now come up
 with a smaller set of library paths for Java on Oracle-format JDK
 (including OpenJDK).  This helps avoid conflicts between libraries
 (such as <code>libjpeg</code>) supplied in the JDK and system
 libraries.  This can always be overridden if needed: see the
 &lsquo;R Installation and Administration&rsquo; manual.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>beta(a, b)</code> could overflow to infinity in its
 calculations when one of <code>a</code> and <code>b</code> was less than one.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=15075">PR#15075</a>)
 </p>
 </li>
 <li> <p><code>lbeta(a, b)</code> no longer gives <code>NaN</code> if <code>a</code> or
 <code>b</code> is very small (in the denormalized range).
 </p>
 </li>
 <li> <p><code>bquote()</code> is now able to substitute default arguments in
 single-argument functions.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=15077">PR#15077</a>)
 </p>
 </li>
 <li> <p><code>browseEnv(html = FALSE)</code> would segfault if called from
 <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> (not <code>R.app</code>) on a CRAN-style Mac OS X build of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 </p>
 </li>
 <li> <p><span class="samp">[[&lt;-</span> for lists (generic vectors) needed to increment
 <code>NAMED</code> count when RHS is used more than once. (<a href="https://bugs.R-project.org/show_bug.cgi?id=15098">PR#15098</a>)
 </p>
 </li>
 <li><p> On Windows, warnings about opening a file or pipe with a
 non-ASCII description were sometimes output in UTF-8 rather than
 in the current locale's character set.
 </p>
 </li>
 <li><p> The <code>call()</code> function did not duplicate its arguments.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=15115">PR#15115</a>)
 </p>
 </li>
 <li> <p><code>TukeyHSD()</code> could give <code>NA</code> results with some
 <span class="samp">na.action</span> methods such as <code>na.exclude()</code>.
 (Hinted at on R-help by John Fox.)
 </p>
 </li>
 <li><p> The deprecated <code>svd(X, LINPACK = TRUE)</code> could alter
 <code>X</code> in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.15.[12].  (Reported by Bill Dunlap.)
 </p>
 </li>
 <li><p> Under Windows, <code>file.link()</code> and <code>file.symlink()</code>
 used the link name twice, so would always fail. (Reported
 by Rui Barradas/Oliver Soong).
 </p>
 </li>
 <li> <p><code>summaryRprof(memory = "both")</code> mixed up the units of
 Vcells and Ncells: it now works in bytes.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=15138">PR#15138</a>)
 </p>
 </li>
 <li> <p><code>tools::Rd2HTML()</code> would sometimes delete text. (<a href="https://bugs.R-project.org/show_bug.cgi?id=15134">PR#15134</a>)
 </p>
 </li>
 <li> <p><code>plot()</code> failed for <code>"table"</code> objects containing
 just one entry.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=15118">PR#15118</a>)
 </p>
 </li>
 <li> <p><code>embedFonts()</code> needed to quote some filepaths.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=15149">PR#15149</a>)
 </p>
 </li>
 <li> <p><code>parallel::mccollect()</code> handled <code>NULL</code> returns
 incorrectly (removing the element rather than setting it to
 <code>NULL</code>).
 </p>
 </li>
 <li><p> The full reference index (&lsquo;<span class="file">fullrefman.pdf</span>&rsquo;) was missing
 packages <span class="pkg">compiler</span> and <span class="pkg">parallel</span>.
 </p>
 </li>
 <li><p> The report for
 <code>optim(method = "L-BFGS-B", control = list(trace = 1))</code>
 reported the last completed and not the current iteration, unlike
 other methods and trace levels.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=15103">PR#15103</a>)
 </p>
 </li>
 <li> <p><code>qt(1e-12, 1.2)</code> no longer gives <code>NaN</code>.
 </p>
 </li>
 <li> <p><code>dt(1e160, 1.2, log=TRUE)</code> no longer gives <code>-Inf</code>.
 </p>
 </li>
 <li><p> On Windows the <code>untar()</code> function now quotes the
 directory name when using an external <code>tar</code> utility, so
 <code>R CMD check</code> will handle pathnames containing spaces.
 </p>
 </li>
 <li><p> The version for Windows 8 and Windows Server 2012 is now
 displayed by <code>win.version()</code>.  (Reported by Gabor
 Grothendieck.)
 </p>
 </li>
 <li><p> The custom Windows installer target <code>myR</code> in the
 installer &lsquo;<span class="file">Makefile</span>&rsquo; did not work in 2.15.2.  (Reported
 by Erich Neuwirth.)
 </p>
 </li>
 <li> <p><code>aperm(matrix(1:6, 2, dimnames=list(A={}, B={})), "A")</code>
 no longer segfaults.
 </p>
 </li>
 <li><p> Expressions involving user defined operators were not
 always deparsed faithfully. (<a href="https://bugs.R-project.org/show_bug.cgi?id=15179">PR#15179</a>)
 </p>
 </li>
 <li><p> The <code>enc2utf8()</code> function converted <code>NA_character_</code>
 to <code>"NA"</code> in non-UTF-8 locales. (<a href="https://bugs.R-project.org/show_bug.cgi?id=15201">PR#15201</a>)
 </p>
 </li>
 <li><p> The <code>exclude</code> argument to <code>xtabs()</code> was ignored
 for <code>"factor"</code> arguments.
 </p>
 </li>
 <li><p> On Windows, work around an event-timing problem when
 the RGui console was closed from the &lsquo;X&rsquo; control and the closure
 cancelled. (This would on some 64-bit systems crash <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>, typically
 those with a slow GPU relative to the CPU.)
 </p>
 </li>
 <li><p> On unix <code>Rscript</code> will pass the <code>r_arch</code> setting
 it was compiled with on to the R process so that the architecture
 of <code>Rscript</code> and that of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> will match unless overridden.
 </p>
 </li>
 <li><p> On Windows, <code>basename()</code>, <code>dirname()</code> and
 <code>file.choose()</code> have more support for long non-ASCII file
 names with 260 or more bytes when expressed in UTF-8.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.15.2</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li><p> The <code>X11()</code> window gains an icon: the
 latter may be especially useful on Ubuntu's &lsquo;Unity&rsquo;
 interface.
 </p>
 <p>The <code>WM_CLASS</code> should be set in circumstances where the
 Window Manager failed to make use of X11 resource settings.
 </p>
 <p>(Contributed by Philip Johnson.)
 </p>
 </li>
 <li><p> The <code>"Date"</code> and <code>"POSIXt"</code> methods for
 <code>cut()</code> will accept an unsorted <code>breaks</code> argument (as
 the default method does, although this was undocumented).
 (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14961">PR#14961</a>.)
 </p>
 </li>
 <li><p> Reference class methods (in the <span class="pkg">methods</span> package) that
 use other methods in an indirect way (e.g., by <code>sapply()</code>)
 must tell the code analysis to include that method.  They can now
 do so by invoking <code>$usingMethods()</code>.
 </p>
 </li>
 <li><p> More Polish translations are available: for the
 <code>RGui</code> menus and for several recommended packages.
 </p>
 </li>
 <li><p> Multistratum MANOVA works. In fact, it seems to have done so
 for years in spite of the help page claiming it did not.
 </p>
 </li>
 <li> <p><code>qqline()</code> has new optional arguments
 <code>distribution</code>, <code>probs</code> and <code>qtype</code>, following the
 example of <a href="https://CRAN.R-project.org/package=lattice"><span class="pkg">lattice</span></a>'s <code>panel.qqmathline()</code>.
 </p>
 </li>
 <li><p> The handling of single quotes in the <code>en@quot</code>
 pseudo-language has been slightly improved.  Double quotes are no
 longer converted.
 </p>
 </li>
 <li><p> New functions <code>checkPoFiles()</code> and <code>checkPoFile()</code>
 have been added to the <span class="pkg">tools</span> package to check for
 consistency of format strings in translation files.
 </p>
 </li>
 <li> <p><code>model.matrix(~1, ...)</code> now also contains the same
 rownames that less trivial formulae produce.  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14992">PR#14992</a>,
 changes the output of several packages.)
 </p>
 </li>
 <li><p> Misuse of <code>rep()</code> on undocumented types of objects
 (e.g., calls) is now reported as an error.
 </p>
 </li>
 <li><p> The included LAPACK has been updated to 3.4.1, with some
 patches from the current SVN sources.  (<em>Inter alia</em>, this
 resolves <a href="https://bugs.R-project.org/show_bug.cgi?id=14692">PR#14692</a>.)
 </p>
 </li>
 <li> <p><code>file.copy(recursive = TRUE)</code> has some additional
 checks on user error leading to attempted infinite recursion (and on
 some platforms to crashing <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>).
 </p>
 </li>
 <li><p> PCRE has been updated to version 8.31, a bug-fix release.
 </p>
 </li>
 <li><p> The included version of <code>liblzma</code> has been updated to
 version 5.0.4, a minor bug-fix release.
 </p>
 </li>
 <li><p> New function <code>.bincode()</code>, a &lsquo;bare-bones&rsquo;
 version of <code>cut.default(labels = FALSE)</code> for use in packages
 with <code>image()</code> methods.
 </p>
 </li>
 <li><p> The HTML manuals now use directional single quotes.
 </p>
 </li>
 <li> <p><code>maintainer()</code> now converts embedded new lines to
 spaces.  It no longer gives a non-obvious error for non-installed
 packages.
 </p>
 </li>
 <li><p> The <code>X11()</code> device has some protection against
 being used with forked processes <em>via</em> package <span class="pkg">parallel</span>.
 </p>
 </li>
 <li><p> Setting the environment variable <span class="env">R_OSX_VALGRIND</span> (to
 any value) allows <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> to be run under <code>valgrind</code> on Mac OS
 10.6 and 10.7 (<code>valgrind</code> currently has very limited
 support for 10.8), provided <code>system()</code> is not used (directly
 or indirectly).  This should not be needed for <code>valgrind</code>
 &gt;= 3.8.1.
 </p>
 </li>
 <li><p> The <code>model.frame()</code> method for <code>"lm"</code> uses
 <code>xlevels</code>: this is safer if <code>data</code> was supplied or
 <code>model = FALSE</code> was used and the levels of factors used in
 the fit had been re-ordered since fitting.
 </p>
 <p>Similarly, <code>model.frame(fm, data=&lt;data&gt;)</code> copies across
 the variables used for safe prediction from the fit.
 </p>
 </li>
 <li><p> Functions such as <code>parLapply()</code> in package
 <span class="pkg">parallel</span> can make use of a default cluster if one is set.
 (Reported by Martin Morgan.)
 </p>
 </li>
 <li> <p><code>chol(pivot = TRUE, LINPACK = FALSE)</code> is now available
 using LAPACK 3.2 subroutine <code>DPSTRF</code>.
 </p>
 </li>
 <li><p> The functions <code>.C()</code>, <code>.Call()</code>,
 <code>.External()</code> and <code>.Fortran()</code> now check that they are
 called with an unnamed first argument: the formal arguments were
 changed from <code>name=</code> to <code>.NAME=</code> in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.0, but some
 packages were still using the old name.  This is currently a
 warning, but will be an error in future.
 </p>
 </li>
 <li> <p><code>step()</code> no longer tries to improve a model with
 AIC of <code>-Inf</code> (a perfect fit).
 </p>
 </li>
 <li> <p><code>spline()</code> and <code>splinefun()</code> gain a new method
 <code>"hyman"</code>, an implementation of Hyman's method of
 constructing monotonic interpolation splines.  (Based on
 contributions of Simon Wood and Rob Hyndman.)
 </p>
 </li>
 <li><p> On Windows, the C stack size has been increased to 64MB (it
 has been 10MB since the days of 32MB RAM systems).
 </p>
 </li></ul>


 <h4>PERFORMANCE IMPROVEMENTS</h4>


 <ul>
 <li> <p><code>array()</code> is now implemented in C code (for speed) when
 <code>data</code> is atomic or an unclassed list (so it is known that
 <code>as.vector(data)</code> will have no class to be used by <code>rep()</code>).
 </p>
 </li>
 <li> <p><code>rep()</code> is faster and uses less memory,
 substantially so in some common cases (e.g., if <code>times</code> is of
 length one or <code>length.out</code> is given, and <code>each = 1</code>).
 </p>
 </li>
 <li> <p><code>findInterval()</code>, <code>tabulate()</code>, <code>cut()</code>,
 <code>hist()</code> and <code>image.default()</code> all use <code>.Call()</code>
 and are more efficient.
 </p>
 </li>
 <li> <p><code>duplicated()</code>, <code>unique()</code> and similar now
 support vectors of lengths above <i>2^29</i> on 64-bit
 platforms.
 </p>
 </li>
 <li><p> Omitting <code>PACKAGE</code> in <code>.C()</code> etc calls was
 supposed to make use of the DLL from the namespace within which
 the enclosing function was defined.  It was less successful in
 doing so than it might be, and gave no indication it had failed.
 </p>
 <p>A new search strategy is very successful and gives a warning
 when it fails.  In most cases this is because the entry point is
 not actually provided by that package (and so <code>PACKAGE</code>
 should be used to indicate which package is intended) but in some
 the namespace does not have a DLL specified by a
 <code>useDynLib()</code> directive so <code>PACKAGE</code> is required.
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li> <p><code>R CMD check</code> now checks if a package can be loaded
 by <code>library(pkgname, lib.loc = "somewhere")</code> without being on
 the library search path (unless it is already installed in
 <code>.Library</code>, when it always will be).
 </p>
 </li>
 <li> <p><code>R CMD check --as-cran</code> notes &lsquo;hidden&rsquo; files
 and directories (with names starting with a dot) that are not
 needed for the operation of <code>R CMD INSTALL</code> or <code>R CMD
 	build</code>: such files should be excluded from the published tarball.
 </p>
 </li>
 <li> <p><code>R CMD check</code> (if checking subdirectories) checks
 that the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> code in any demos is ASCII and can be parsed, and
 warns if not.
 </p>
 </li>
 <li><p> When <code>R CMD Rd2pdf</code> is used with &lsquo;<span class="file">inputenx.sty</span>&rsquo;,
 it allows further characters (mainly for Eastern European
 languages) by including &lsquo;<span class="file">ix-utf8enc.dfu</span>&rsquo; (if available).
 (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14989">PR#14989</a>.)
 </p>
 </li>
 <li> <p><code>R CMD build</code> now omits several types of hidden
 files/directories, including &lsquo;<span class="file">inst/doc/.Rinstignore</span>&rsquo;,
 &lsquo;<span class="file">vignettes/.Rinstignore</span>&rsquo;, (&lsquo;<span class="file">.Rinstignore</span>&rsquo; should be at
 top level), &lsquo;<span class="file">.deps</span>&rsquo; under &lsquo;<span class="file">src</span>&rsquo;, &lsquo;<span class="file">.Renviron</span>&rsquo;,
 &lsquo;<span class="file">.Rprofile</span>&rsquo;, &lsquo;<span class="file">.Rproj.user</span>&rsquo;, &lsquo;<span class="file">.backups</span>&rsquo;,
 &lsquo;<span class="file">.cvsignore</span>&rsquo;, &lsquo;<span class="file">.cproject</span>&rsquo;, &lsquo;<span class="file">.directory</span>&rsquo;,
 &lsquo;<span class="file">.dropbox</span>&rsquo;, &lsquo;<span class="file">.exrc</span>&rsquo;, &lsquo;<span class="file">.gdb.history</span>&rsquo;,
 &lsquo;<span class="file">.gitattributes</span>&rsquo;, &lsquo;<span class="file">.gitignore</span>&rsquo;, &lsquo;<span class="file">.gitmodules</span>&rsquo;,
 &lsquo;<span class="file">.hgignore</span>&rsquo;, &lsquo;<span class="file">.hgtags</span>&rsquo;, &lsquo;<span class="file">.htaccess</span>&rsquo;,
 &lsquo;<span class="file">.latex2html-init</span>&rsquo;, &lsquo;<span class="file">.project</span>&rsquo;, &lsquo;<span class="file">.seed</span>&rsquo;,
 &lsquo;<span class="file">.settings</span>&rsquo;, &lsquo;<span class="file">.tm_properties</span>&rsquo; and various leftovers.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now checks for <code>.C()</code>,
 <code>.Call()</code>, <code>.External()</code> and <code>.Fortran()</code> calls in
 other packages, and gives a warning on those found from <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> itself
 (which are not part of the API and change without notice: many
 will be changed for <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 3.0.0).
 </p>
 </li></ul>


 <h4>C-LEVEL FACILITIES</h4>


 <ul>
 <li><p> The limit for <code>R_alloc</code> on 64-bit platforms has been
 raised to just under 32GB (from just under 16GB).
 </p>
 </li>
 <li><p> The misuse of <code>.C("name", ..., PACKAGE = foo)</code> where
 <code>foo</code> is an arbitrary <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> object is now an error.
 </p>
 <p>The misuse <code>.C("name",..., PACKAGE = "")</code> is now warned about
 in <code>R CMD check</code>, and will be an error in future.
 </p>
 </li></ul>


 <h4>DEPRECATED AND DEFUNCT</h4>


 <ul>
 <li><p> Use of <code>array()</code> with a 0-length <code>dim</code> argument is
 deprecated with a warning (and was contrary to the documentation).
 </p>
 </li>
 <li><p> Use of <code>tapply()</code> with a 0-length <code>INDEX</code> list is
 deprecated with a warning.
 </p>
 </li>
 <li> <p><span class="samp">Translation</span> packages are deprecated.
 </p>
 </li>
 <li><p> Calling <code>rep()</code> or <code>rep.int()</code> on a pairlist is
 deprecated and will give a warning.  In any case, <code>rep()</code>
 converted a pairlist to a list so you may as well do that
 explicitly.
 </p>
 </li>
 <li><p> Entry point <code>rcont2</code> is no longer part of the API, and
 will move to package <span class="pkg">stats</span> in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 3.0.0.
 </p>
 </li>
 <li><p> The &lsquo;internal&rsquo; graphics device invoked by
 <code>.Call("R_GD_nullDevice", package = "grDevices")</code> is about to
 be removed: use <code>pdf(file = NULL)</code> instead.
 </p>
 </li>
 <li> <p><code>eigen(EISPACK = TRUE)</code>,
 <code>chol(pivot = FALSE, LINPACK = TRUE)</code>,
 <code>chol2inv(LINPACK = TRUE)</code>, <code>solve(LINPACK = TRUE)</code> and
 <code>svd(LINPACK = TRUE)</code> are deprecated and give a warning.
 </p>
 <p>They were provided for compatibility with <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 1.7.0 (Mar 2003)!
 </p>
 </li>
 <li><p> The &lsquo;internal function&rsquo; <code>kappa.tri()</code> has been
 renamed to <code>.kappa_tri()</code> so it is not inadvertently called
 as a method for class <code>"tri"</code>.
 </p>
 </li>
 <li><p> Functions <code>sessionData()</code> and <code>browseAll()</code> in
 package <span class="pkg">methods</span> are on a help page describing them as
 &lsquo;deprecated&rsquo; and are now formally deprecated.
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li><p> For a Windows or Mac OS X binary package install,
 <code>install.packages()</code> will check if a source package is
 available on the same repositories, and report if it is a later
 version or there is a source package but no binary package
 available.
 </p>
 <p>This check can be suppressed: see the help page.
 </p>
 </li>
 <li> <p><code>install.packages(type = "both")</code> has been enhanced.
 In interactive use it will ask whether to choose the source
 version of a package if the binary version is older and contains
 compiled code, and also asks if source packages with no binary
 version should be installed).
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> There is a new <code>configure</code> option
 <span class="option">--with-libtiff</span> (mainly in case the system installation
 needs to be avoided).
 </p>
 </li>
 <li><p> LAPACK 3.4.1 does use some Fortran 90 features, so
 <code>g77</code> no longer suffices.
 </p>
 </li>
 <li><p> If an external LAPACK is used, it must be version 3.2 or later.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li><p> On Windows, starting <code>Rterm</code> via <code>R.exe</code>
 caused Ctrl-C to misbehave.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14948">PR#14948</a>)
 </p>
 </li>
 <li><p> The <code>tools::latexToUtf8()</code> function missed conversions that
 were contained within braces.
 </p>
 </li>
 <li><p> Long timezone specifications (such as a file name preceded
 by <code>:</code>) could crash <code>as.POSIXlt</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14945">PR#14945</a>)
 </p>
 </li>
 <li> <p><code>R CMD build --resave-data</code> could fail if there was no
 &lsquo;<span class="file">data</span>&rsquo; directory but there was an &lsquo;<span class="file">R/sysdata.rda</span>&rsquo; file.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14947">PR#14947</a>)
 </p>
 </li>
 <li> <p><code>is.na()</code> misbehaved on a 0-column data frame.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14959">PR#14959</a>)
 </p>
 </li>
 <li> <p><code>anova.lmlist()</code> failed if <code>test</code> was
 supplied.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14960">PR#14960</a>)
 </p>
 <p>It was unable to compute Cp tests for object of class <code>"lm"</code>
 (it assumed class <code>"glm"</code>).
 </p>
 </li>
 <li><p> The <code>formula</code> method for <code>sunflowerplot()</code> now
 allows <code>xlab</code> and <code>ylab</code> to be set.  (Reported by Gerrit
 Eichner.)
 </p>
 </li>
 <li><p> The <code>"POSIXt"</code> and <code>"Date"</code> methods for <code>hist()</code>
 could fail on Windows where adjustments to the right-hand boundary
 crossed a DST transition time.
 </p>
 </li>
 <li><p> On Windows, the code in <code>as.POSIXct()</code> to handle
 incorrectly specified <code>isdst</code> fields might have resulted in
 <code>NA</code> being returned.
 </p>
 </li>
 <li> <p><code>aov()</code> and <code>manova()</code> gave spurious warning about
 singular error model in the multiresponse case.
 </p>
 </li>
 <li><p> In <code>ns()</code> and <code>bs()</code>, specifying <code>knots = NULL</code>
 is now equivalent to omitting it, also when <code>df</code> is
 specified.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14970">PR#14970</a>)
 </p>
 </li>
 <li> <p><code>sprintf()</code> did not accept numbered arguments ending
 in zero.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14975">PR#14975</a>)
 </p>
 </li>
 <li> <p><code>rWishart()</code> could overflow the C stack and maybe crash
 the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> process for dimensions of several hundreds or more.
 (Reported by Michael Braun on R-sig-mac.)
 </p>
 </li>
 <li><p> Base package vignettes (e.g., <code>vignette("Sweave")</code>) were
 not fully installed in builds of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> from the tarball.
 </p>
 </li>
 <li> <p><code>lchoose()</code> and <code>choose()</code> could overflow the C
 stack and crash <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 </p>
 </li>
 <li><p> When given a 0-byte file and asked to keep source
 references, <code>parse()</code> read input from <code>stdin()</code>
 instead.
 </p>
 </li>
 <li> <p><code>pdf(compress = TRUE)</code> did not delete temporary files
 it created until the end of the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> session.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14991">PR#14991</a>)
 </p>
 </li>
 <li> <p><code>logLik()</code> did not detect the error of applying it
 to a multiple-response linear model.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=15000">PR#15000</a>)
 </p>
 </li>
 <li> <p><code>file.copy(recursive = TRUE)</code> did not always report
 <code>FALSE</code> for a failure two or more directories deep.
 </p>
 </li>
 <li> <p><code>qgeom()</code> could return <code>-1</code> for extremely small
 <code>q</code>. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14967">PR#14967</a>)
 </p>
 </li>
 <li> <p><code>smooth.spline()</code> used <code>DUP = FALSE</code> which allowed
 its compiled C code to change the function: this was masked by the
 default byte-compilation. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14965">PR#14965</a>)
 </p>
 </li>
 <li><p> In Windows, the GUI preferences for foreground color were
 not always respected.  (Reported by Benjamin Wells.)
 </p>
 </li>
 <li><p> On OS X, the Quartz versions of the bitmap devices did not
 respect <code>antialias = "none"</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=15006">PR#15006</a>)
 </p>
 </li>
 <li> <p><code>unique()</code> and similar would infinite-loop if called on
 a vector of length &gt; <i>2^29</i> (but reported that the
 vector was too long for <i>2^30</i> or more).
 </p>
 </li>
 <li> <p><code>parallel::stopCluster()</code> now works with MPI
 clusters without <a href="https://CRAN.R-project.org/package=snow"><span class="pkg">snow</span></a> being on the search path.
 </p>
 </li>
 <li> <p><code>terms.formula()</code> could exhaust the stack, and the
 stack check did not always catch this before the segfault.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=15013">PR#15013</a>)
 </p>
 </li>
 <li> <p><code>sort.list(method = "radix")</code> could give incorrect
 results on certain compilers (seen with <code>clang</code> on Mac OS
 10.7 and <code>Xcode 4.4.1</code>).
 </p>
 </li>
 <li> <p><code>backsolve(T, b)</code> gave incorrect results when
 <code>nrows(b) &gt; ncols(T)</code> and <code>b</code> had more than one column.
 </p>
 <p>It could segfault or give nonsense if <code>k</code> was specified as
 more than <code>ncols(T)</code>.
 </p>
 </li>
 <li> <p><code>smooth.spline()</code> did not check that a specified
 numeric <code>spar</code> was of length 1, and gave corrupt results if
 it was of length 0.
 </p>
 </li>
 <li><p> Protection added to <code>do_system</code>. (<a href="https://bugs.R-project.org/show_bug.cgi?id=15025">PR#15025</a>)
 </p>
 </li>
 <li><p> Printing of vectors with names &gt; 1000 characters now works
 correctly rather than truncating. (<a href="https://bugs.R-project.org/show_bug.cgi?id=15028">PR#15028</a>)
 </p>
 </li>
 <li> <p><code>qr()</code> for a complex matrix did not pivot the column names.
 </p>
 </li>
 <li> <p><code>--with-blas='-framework vecLib'</code> now also works on OS X 10.8.
 </p>
 </li>
 <li> <p><code>R CMD check</code> no longer fails with an error if
 a &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file incorrectly contains a blank line.
 (Reported by Bill Dunlap.)
 </p>
 </li>
 <li> <p><code>install.packages(type = "both")</code> could call
 <code>chooseCRANmirror()</code> twice.
 </p>
 </li>
 <li> <p><code>lm.wfit()</code> could segfault in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.15.1 if all the
 weights were zero. (<a href="https://bugs.R-project.org/show_bug.cgi?id=15044">PR#15044</a>)
 </p>
 </li>
 <li><p> A malformed package name could cause <code>R CMD INSTALL</code>
 to write outside the target library.
 </p>
 </li>
 <li><p> Some of the quality control functions
 (e.g., <code>tools::checkFF()</code>) were wrongly identifying the source
 of S4 methods in a package and so not checking them.
 </p>
 </li>
 <li><p> The default type of display by <code>browseEnv()</code> when using
 <code>R.app</code> on Mac OS X has been incorrect for a long time.
 </p>
 </li>
 <li><p> The implementation of <code>importMethodsFrom</code> in a
 <code>NAMESPACE</code> file could be confused and fail to find generics
 when importing from multiple packages (reported and fixed by
 Michael Lawrence).
 </p>
 </li>
 <li><p> The detection of the C stack direction is better protected
 against compiler optimization. (<a href="https://bugs.R-project.org/show_bug.cgi?id=15011">PR#15011</a>)
 </p>
 </li>
 <li><p> Long custom line types would sometimes segfault on the
 cairographics-based devices. (<a href="https://bugs.R-project.org/show_bug.cgi?id=15055">PR#15055</a>)
 </p>
 </li>
 <li> <p><code>tools::checkPoFile()</code> unprotected too early in its C
 code and so segfaulted from time to time.
 </p>
 </li>
 <li><p> The Fortran code underlying <code>nlminb()</code> could
 infinite-loop if any of the input functions returned <code>NA</code> or
 <code>NaN</code>.  This is now an error for the gradient or Hessian, and
 a warning for the function (with the value replaced by
 <code>Inf</code>).  (In part, <a href="https://bugs.R-project.org/show_bug.cgi?id=15052">PR#15052</a>.)
 </p>
 </li>
 <li><p> The code for creating <code>coerce()</code> methods could
 generate false notes about ambiguous selection; the notes have
 been suppressed for this function.
 </p>
 </li>
 <li> <p><code>arima.sim()</code> could give too long an output in some
 corner cases (in part, <a href="https://bugs.R-project.org/show_bug.cgi?id=15068">PR#15068</a>).
 </p>
 </li>
 <li> <p><code>anova.glm()</code> with <code>test = "Rao"</code> didn't work when
 models included an offset. (Reported by Søren Feodor Nielsen.)
 </p>
 </li>
 <li> <p><code>as.data.frame.matrix()</code> could return invalid data
 frame with no <code>row.names</code> attribute for 0-row matrix. (Reported
 by Hervé Pagès.)
 </p>
 </li>
 <li><p> Compilation with the <code>vecLib</code> or <code>Accelerate</code>
 frameworks on OS X without using that also for LAPACK is more
 likely to be successful.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.15.1</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>source()</code> now uses <code>withVisible()</code> rather than
 <code>.Internal(eval.with.vis)</code>.  This sometimes alters tracebacks
 slightly.
 </p>
 </li>
 <li> <p><code>install.packages("pkg_version.tgz")</code> on Mac OS X now
 has sanity checks that this is actually a binary package (as
 people have tried it with incorrectly named source packages).
 </p>
 </li>
 <li> <p><code>splineDesign()</code> and <code>spline.des()</code> in package
 <span class="pkg">splines</span> have a new option <code>sparse</code> which can be used
 for efficient construction of a sparse B-spline design matrix
 (<em>via</em> <a href="https://CRAN.R-project.org/package=Matrix"><span class="pkg">Matrix</span></a>).
 </p>
 </li>
 <li> <p><code>norm()</code> now allows <code>type = "2"</code> (the
 &lsquo;spectral&rsquo; or 2-norm) as well, mainly for didactical
 completeness.
 </p>
 </li>
 <li> <p><code>pmin()</code> and <code>pmax())</code> now also work when one of
 the inputs is of length zero and others are not, returning a
 zero-length vector, analogously to, say, <code>+</code>.
 </p>
 </li>
 <li> <p><code>colorRamp()</code> (and hence <code>colorRampPalette()</code>) now
 also works for the boundary case of just one color when the ramp is
 flat.
 </p>
 </li>
 <li> <p><code>.C()</code> gains some protection against the misuse of
 character vector arguments.  (An all too common error is to pass
 <code>character(N)</code>, which initializes the elements to <code>""</code>,
 and then attempt to edit the strings in-place, sometimes forgetting
 to terminate them.)
 </p>
 </li>
 <li><p> Calls to the new function <code>globalVariables()</code> in
 package <span class="pkg">utils</span> declare that functions and other objects in a
 package should be treated as globally defined, so that <code>CMD
       check</code> will not note them.
 </p>
 </li>
 <li> <p><code>print(packageDescription(*))</code> trims the <code>Collate</code>
 field by default.
 </p>
 </li>
 <li><p> The included copy of <code>zlib</code> has been updated to version
 1.2.7.
 </p>
 </li>
 <li><p> A new option <code>"show.error.locations"</code> has been added.
 When set to <code>TRUE</code>, error messages will contain the
 location of the most recent call containing source reference
 information. (Other values are supported as well; see
 <code>?options</code>.)
 </p>
 </li>
 <li><p> The NA warning messages from e.g. <code>pchisq()</code> now report
 the call to the closure and not that of the <code>.Internal</code>.
 </p>
 </li>
 <li><p> Added Polish translations by Łukasz Daniel.
 </p>
 </li></ul>


 <h4>PERFORMANCE IMPROVEMENTS</h4>


 <ul>
 <li><p> In package <span class="pkg">parallel</span>, <code>makeForkCluster()</code> and the
 multicore-based functions use native byte-order for serialization
 (deferred from 2.15.0).
 </p>
 </li>
 <li> <p><code>lm.fit()</code>, <code>lm.wfit()</code>, <code>glm.fit()</code> and
 <code>lsfit()</code> do less copying of objects, mainly by using
 <code>.Call()</code> rather than <code>.Fortran()</code>.
 </p>
 </li>
 <li> <p><code>.C()</code> and <code>.Fortran()</code> do less copying: arguments
 which are raw, logical, integer, real or complex vectors and are
 unnamed are not copied before the call, and (named or not) are not
 copied after the call.  Lists are no longer copied (they are
 supposed to be used read-only in the C code).
 </p>
 </li>
 <li> <p><code>tabulate()</code> makes use of <code>.C(DUP = FALSE)</code> and
 hence does not copy <code>bin</code>.  (Suggested by Tim Hesterberg.)
 It also avoids making a copy of a factor argument <code>bin</code>.
 </p>
 </li>
 <li><p> Other functions (often or always) doing less copying include
 <code>cut()</code>, <code>dist()</code>, the complex case of <code>eigen()</code>,
 <code>hclust()</code>, <code>image()</code>, <code>kmeans()</code>, <code>loess()</code>,
 <code>stl()</code> and <code>svd(LINPACK = TRUE)</code>.
 </p>
 </li>
 <li><p> There is less copying when using primitive replacement
 functions such as <code>names()</code>, <code>attr()</code> and
 <code>attributes()</code>.
 </p>
 </li></ul>


 <h4>DEPRECATED AND DEFUNCT</h4>


 <ul>
 <li><p> The converters for use with <code>.C()</code> (see
 <code>?getCConverterDescriptions</code>) are deprecated: use the
 <code>.Call()</code> interface instead.  There are no known examples
 (they were never fully documented).
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li><p> For <code>R CMD check</code>, a few people have reported
 problems with junctions on Windows (although they were tested on
 Windows 7, XP and Server 2008 machines and it is unknown under
 what circumstances the problems occur).  Setting the environment
 variable <span class="env">R_WIN_NO_JUNCTIONS</span> to a non-empty value (e.g., in
 &lsquo;<span class="file">~/.R/check.Renviron</span>&rsquo;) will force copies to be used instead.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li> <p><code>R CMD INSTALL</code> with <span class="env">_R_CHECK_INSTALL_DEPENDS_</span>
 set to a true value (as done by <code>R CMD check --as-cran</code>)
 now restricts the packages available when lazy-loading as well
 as when test-loading (since packages such as <a href="https://CRAN.R-project.org/package=ETLUtils"><span class="pkg">ETLUtils</span></a>
 and <a href="https://CRAN.R-project.org/package=agsemisc"><span class="pkg">agsemisc</span></a> had top-level calls to <code>library()</code>
 for undeclared packages).
 </p>
 <p>This check is now also available on Windows.
 </p>
 </li></ul>


 <h4>C-LEVEL FACILITIES</h4>


 <ul>
 <li><p> C entry points <code>mkChar</code> and <code>mkCharCE</code> now check
 that the length of the string they are passed does not exceed
 <i>2^31-1</i> bytes: they used to overflow with
 unpredictable consequences.
 </p>
 </li>
 <li><p> C entry points <code>R_GetCurrentSrcref</code> and
 <code>R_GetSrcFilename</code> have been added to the API to allow
 debuggers access to the source references on the stack.
 </p>
 </li></ul>


 <h4>WINDOWS-SPECIFIC CHANGES</h4>


 <ul>
 <li><p> Windows-specific changes will now be announced in this file
 (&lsquo;<span class="file">NEWS</span>&rsquo;).  Changes up and including <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.15.0 remain in the
 &lsquo;<span class="file">CHANGES</span>&rsquo; file.
 </p>
 </li>
 <li><p> There are two new environment variables which
 control the defaults for command-line options.
 </p>
 <p>If <span class="env">R_WIN_INTERNET2</span> is set to a non-empty value, it is as if
 <span class="option">--internet2</span> was used.
 </p>
 <p>If <span class="env">R_MAX_MEM_SIZE</span> is set, it gives the default memory limit
 if <span class="option">--max-mem-size</span> is not specified: invalid values being
 ignored.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>lsfit()</code> lost the names from the residuals.
 </p>
 </li>
 <li><p> More cases in which <code>merge()</code> could create a data frame
 with duplicate column names now give warnings.  Cases where
 names specified in <code>by</code> match multiple columns are errors.
 </p>
 </li>
 <li><p> Nonsense uses such as <code>seq(1:50, by = 5)</code> (from package
 <a href="https://CRAN.R-project.org/package=plotrix"><span class="pkg">plotrix</span></a>) and <code>seq.int(1:50, by = 5)</code> are now errors.
 </p>
 </li>
 <li><p> The residuals in the 5-number summary printed by
 <code>summary()</code> on an <code>"lm"</code> object are now explicitly
 labelled as weighted residuals when non-constant weights are
 present.  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14840">PR#14840</a>.)
 </p>
 </li>
 <li> <p><code>tracemem()</code> reported that all objects were copied
 by <code>.C()</code> or <code>.Fortran()</code> whereas only some object types
 were ever copied.
 </p>
 <p>It also reported and marked as copies <em>some</em> transformations
 such as <code>rexp(n, x)</code>: it no longer does so.
 </p>
 </li>
 <li><p> The <code>plot()</code> method for class <code>"stepfun"</code> only
 used the optional <code>xval</code> argument to compute <code>xlim</code> and
 not the points at which to plot (as documented).  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14864">PR#14864</a>)
 </p>
 </li>
 <li><p> Names containing characters which need to be escaped were
 not deparsed properly.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14846">PR#14846</a>)
 </p>
 </li>
 <li><p> Trying to update (recommended) packages in
 &lsquo;<span class="file">R_HOME/library</span>&rsquo; without write access is now dealt with more
 gracefully.  Further, such package updates may be skipped (with a
 warning), when a newer installed version is already going to be
 used from <code>.libPaths()</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14866">PR#14866</a>)
 </p>
 </li>
 <li> <p><code>hclust()</code> is now fast again (as up to end of 2003),
 with a different fix for the &quot;median&quot;/&quot;centroid&quot; problem.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=4195">PR#4195</a>).
 </p>
 </li>
 <li> <p><code>get_all_vars()</code> failed when the data came entirely
 from vectors in the global environment. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14847">PR#14847</a>)
 </p>
 </li>
 <li> <p><code>R CMD check</code> with <span class="env">_R_CHECK_NO_RECOMMENDED_</span>
 set to a true value (as done by the <code>--as-cran</code> option)
 could issue false errors if there was an indirect dependency
 on a recommended package.
 </p>
 </li>
 <li> <p><code>formatC()</code> uses the C entry point <code>str_signif</code>
 which could write beyond the length allocated for the output string.
 </p>
 </li>
 <li><p> Missing default argument added to implicit S4 generic for
 <code>backsolve()</code>. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14883">PR#14883</a>)
 </p>
 </li>
 <li><p> Some bugs have been fixed in handling load actions that
 could fail to export assigned items or generate spurious warnings
 in <code>CMD check</code> on loading.
 </p>
 </li>
 <li><p> For <code>tiff(type = "windows")</code>, the numbering of per-page
 files except the last was off by one.
 </p>
 </li>
 <li><p> On Windows, loading package <span class="pkg">stats</span> (which is done for
 a default session) would switch line endings on &lsquo;<span class="file">stdout</span>&rsquo;
 and &lsquo;<span class="file">stderr</span>&rsquo; from CRLF to LF.  This affected
 <code>Rterm</code> and <code>R CMD BATCH</code>.
 </p>
 </li>
 <li><p> On Windows, the compatibility function <code>x11()</code> had
 not kept up with changes to <code>windows()</code>, and issued
 warnings about bad parameters. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14880">PR#14880</a>)
 </p>
 </li>
 <li><p> On Windows, the <code>Sys.glob()</code> function did not handle
 UNC paths as it was designed to try to do. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14884">PR#14884</a>)
 </p>
 </li>
 <li><p> In package <span class="pkg">parallel</span>, <code>clusterApply()</code> and similar
 failed to handle a (pretty pointless) length-1 argument. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14898">PR#14898</a>)
 </p>
 </li>
 <li><p> Quartz Cocoa display reacted asynchronously to
 <code>dev.flush()</code> which means that the redraw could be performed
 after the plot has been already modified by subsequent code. The
 redraw is now done synchronously in <code>dev.flush()</code> to allow
 animations without sleep cycles.
 </p>
 </li>
 <li><p> Source locations reported in <code>traceback()</code> were
 incorrect when byte-compiled code was on the stack.
 </p>
 </li>
 <li> <p><code>plogis(x, lower = FALSE, log.p = TRUE)</code> no longer
 underflows early for large x (e.g., 800).
 </p>
 </li>
 <li> <p><code>?Arithmetic</code>'s &ldquo;<code>1 ^ y</code> and <code>y ^ 0</code>
 are <code>1</code>, <em>always</em>&rdquo; now also applies for <code>integer</code>
 vectors <code>y</code>.
 </p>
 </li>
 <li><p> X11-based pixmap devices like <code>png(type = "Xlib")</code> were
 trying to set the cursor style, which triggered some warnings and
 hangs.
 </p>
 </li>
 <li><p> Code executed by the built-in HTTP server no longer allows
 other HTTP clients to re-enter <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> until the current worker
 evaluation finishes, to prevent cascades.
 </p>
 </li>
 <li><p> The <code>plot()</code> and <code>Axis()</code> methods for class
 <code>"table"</code> now respect graphical parameters such as
 <code>cex.axis</code>.  (Reported by Martin Becker.)
 </p>
 </li>
 <li><p> Under some circumstances <code>package.skeleton()</code> would
 give out progress reports that could not be translated and so were
 displayed by question marks.  Now they are always in English.
 (This was seen for CJK locales on Windows, but may have occurred
 elsewhere.)
 </p>
 </li>
 <li><p> The evaluator now keeps track of source references outside
 of functions, e.g. when <code>source()</code> executes a script.
 </p>
 </li>
 <li><p> The replacement method for <code>window()</code> now works
 correctly for multiple time series of class <code>"mts"</code>.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14925">PR#14925</a>)
 </p>
 </li>
 <li> <p><code>is.unsorted()</code> gave incorrect results on non-atomic
 objects such as data frames.  (Reported by Matthew Dowle.)
 </p>
 </li>
 <li><p> The value returned by <code>tools::psnice()</code> for invalid
 <code>pid</code> values was not always <code>NA</code> as documented.
 </p>
 </li>
 <li><p> Closing an <code>X11()</code> window while <code>locator()</code> was
 active could abort the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> process.
 </p>
 </li>
 <li> <p><code>getMethod(f, sig)</code> produced an incorrect error message
 in some cases when <code>f</code> was not a string).
 </p>
 </li>
 <li><p> Using a string as a &ldquo;call&rdquo; in an error condition
 with <code>options(showErrorCalls=TRUE)</code> could cause a segfault.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14931">PR#14931</a>)
 </p>
 </li>
 <li><p> The string <code>"infinity"</code> allowed by C99 was not accepted
 as a numerical string value by e.g. <code>scan()</code> and
 <code>as.numeric()</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14933">PR#14933</a>)
 </p>
 </li>
 <li><p> In <code>legend()</code>, setting some entries of <code>lwd</code> to
 <code>NA</code> was inconsistent (depending on the graphics device) in
 whether it would suppress those lines; now it consistently does
 so.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14926">PR#14926</a>)
 </p>
 </li>
 <li> <p><code>by()</code> failed for a zero-row data frame.  (Reported by
 Weiqiang Qian.)
 </p>
 </li>
 <li><p> The Yates' correction in <code>chisq.test()</code> could be bigger
 than the terms it corrected, previously leading to an infinite
 test statistic in some corner cases which are now reported as
 <code>NaN</code>.
 </p>
 </li>
 <li> <p><code>xgettext()</code> and related functions sometimes returned
 items that were not strings for translation. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14935">PR#14935</a>)
 </p>
 </li>
 <li> <p><code>plot(&lt;lm&gt;, which=5)</code> now correctly labels the factor
 level combinations for the special case where all
 <i>h[i,i]</i> are the same. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14837">PR#14837</a>)
 </p>
 </li>
 <li> <p><code>Sys.glob()</code> caused a segfault if the first element of
 <code>path</code> was <code>NA_character</code>. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14990">PR#14990</a>)
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.15.0</h3>


 <h4>SIGNIFICANT USER-VISIBLE CHANGES</h4>


 <ul>
 <li><p> The behaviour of <code>unlink(recursive = TRUE)</code> for a
 symbolic link to a directory has changed: it now removes the link
 rather than the directory contents (just as <code>rm -r</code> does).
 </p>
 <p>On Windows it no longer follows reparse points (including
 junctions and symbolic links).
 </p>
 </li></ul>


 <h4>NEW FEATURES</h4>


 <ul>
 <li><p> Environment variable <span class="env">RD2DVI_INPUTENC</span> has been renamed
 to <span class="env">RD2PDF_INPUTENC</span>.
 </p>
 </li>
 <li> <p><code>.Deprecated()</code> becomes a bit more flexible, getting an
 <code>old</code> argument.
 </p>
 </li>
 <li><p> Even data-only packages without <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> code need a namespace and
 so may need to be installed under <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.14.0 or later.
 </p>
 </li>
 <li> <p><code>assignInNamespace()</code> has further restrictions on use
 apart from at top-level, as its help page has warned.
 Expect it to be disabled from programmatic use in the future.
 </p>
 </li>
 <li> <p><code>system()</code> and <code>system2()</code> when capturing output
 report a non-zero status in the new <code>"status"</code> attribute.
 </p>
 </li>
 <li> <p><code>kronecker()</code> now has an S4 generic in package
 <span class="pkg">methods</span> on which packages can set methods.  It will be
 invoked by <code>X %x% Y</code> if either <code>X</code> or <code>Y</code> is an S4
 object.
 </p>
 </li>
 <li> <p><code>pdf()</code> accepts forms like <code>file = "|lpr"</code> in the
 same way as <code>postscript()</code>.
 </p>
 </li>
 <li> <p><code>pdf()</code> accepts <code>file = NULL</code>.
 This means that the device does NOT create a PDF file
 (but it can still be queried, e.g., for font metric info).
 </p>
 </li>
 <li> <p><code>format()</code> (and hence <code>print()</code>) on
 <code>"bibentry"</code> objects now uses <code>options("width")</code> to
 set the output width.
 </p>
 </li>
 <li> <p><code>legend()</code> gains a <code>text.font</code> argument. (Suggested
 by Tim Paine, <a href="https://bugs.R-project.org/show_bug.cgi?id=14719">PR#14719</a>.)
 </p>
 </li>
 <li> <p><code>nchar()</code> and <code>nzchar()</code> no longer accept factors
 (as integer vectors).  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=6899">PR#6899</a>.)
 </p>
 </li>
 <li> <p><code>summary()</code> behaves slightly differently (or more
 precisely, its <code>print()</code> method does).  For numeric inputs, the
 number of NAs is printed as an integer and not a real.  For dates
 and datetimes, the number of NAs is included in the printed output
 (the latter being the wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14720">PR#14720</a>).
 </p>
 <p>The <code>"data.frame"</code> method is more consistent with the default
 method: in particular it now applies <code>zapsmall()</code> to
 numeric/complex summaries.
 </p>
 </li>
 <li><p> The number of items retained with <code>options(warn = 0)</code>
 can be set by <code>options(nwarnings=)</code>.
 </p>
 </li>
 <li><p> There is a new function <code>assignInMyNamespace()</code> which
 uses the namespace of the function it is called from.
 </p>
 </li>
 <li> <p><code>attach()</code> allows the default name for an attached file
 to be overridden.
 </p>
 </li>
 <li> <p><code>bxp()</code>, the work horse of <code>boxplot()</code>, now uses a
 more sensible default <code>xlim</code> in the case where <code>at</code> is
 specified differently from <code>1:n</code>, see the discussion on R-devel,
 <a href="https://stat.ethz.ch/pipermail/r-devel/2011-November/062586.html">https://stat.ethz.ch/pipermail/r-devel/2011-November/062586.html</a>.
 </p>
 </li>
 <li><p> New function <code>paste0()</code>, an efficient version of
 <code>paste(*, sep="")</code>, to be used in many places for more
 concise (and slightly more efficient) code.
 </p>
 </li>
 <li><p> Function <code>setClass()</code> in package <span class="pkg">methods</span> now
 returns, invisibly, a generator function for the new class,
 slightly preferred to calling <code>new()</code>, as explained on the
 <code>setClass</code> help page.
 </p>
 </li>
 <li><p> The <code>"dendrogram"</code> method of <code>str()</code> now takes
 its default for <code>last.str</code> from option <code>str.dendrogram.last</code>.
 </p>
 </li>
 <li><p> New simple <code>fitted()</code> method for <code>"kmeans"</code> objects.
 </p>
 </li>
 <li><p> The <code>traceback()</code> function can now be called with
 an integer argument, to display a current stack trace. (Wish of
 <a href="https://bugs.R-project.org/show_bug.cgi?id=14770">PR#14770</a>.)
 </p>
 </li>
 <li> <p><code>setGeneric()</code> calls can be simplified when creating
 a new generic function by supplying the default method as the
 <code>def</code> argument.  See <code>?setGeneric</code>.
 </p>
 </li>
 <li> <p><code>serialize()</code> has a new option <code>xdr = FALSE</code> which
 will use the native byte-order for binary serializations.  In
 scenarios where only little-endian machines are involved (these
 days, close to universal) and (un)serialization takes an
 appreciable amount of time this may speed up noticeably
 transferring data between systems.
 </p>
 </li>
 <li><p> The internal (un)serialization code is faster for long
 vectors, particularly with XDR on some platforms.  (Based on a
 suggested patch by Michael Spiegel.)

 </p>
 </li>
 <li><p> For consistency, circles with zero radius are omitted by
 <code>points()</code> and <code>grid.circle()</code>.  Previously this was
 device-dependent, but they were usually invisible.
 </p>
 </li>
 <li> <p><code>NROW(x)</code> and <code>NCOL(x)</code> now work whenever
 <code>dim(x)</code> looks appropriate, e.g., also for more generalized
 matrices.
 </p>
 </li>
 <li><p> PCRE has been updated to version 8.30.
 </p>
 </li>
 <li><p> The internal <code>R_Srcref</code> variable is now updated
 before the browser stops on entering a function.  (Suggestion of
 <a href="https://bugs.R-project.org/show_bug.cgi?id=14818">PR#14818</a>.)
 </p>
 </li>
 <li><p> There are &lsquo;bare-bones&rsquo; functions <code>.colSums()</code>,
 <code>.rowSums()</code>, <code>.colMeans()</code> and <code>.rowMeans()</code> for
 use in programming where ultimate speed is required.
 </p>
 </li>
 <li><p> The formerly internal function
 <code>.package_dependencies()</code> from package <span class="pkg">tools</span> for
 calculating (recursive) (reverse) dependencies on package databases
 has been renamed to <code>package_dependencies()</code> and is now
 exported.
 </p>
 </li>
 <li><p> There is a new function <code>optimHess()</code> to compute the
 (approximate) Hessian for an <code>optim()</code> solution if
 <code>hessian = TRUE</code> was forgotten.
 </p>
 </li>
 <li> <p><code>.filled.contour()</code> is a &lsquo;bare-bones&rsquo; function to
 add a filled-contour rectangular plot to an already prepared plot
 region.
 </p>
 </li>
 <li><p> The stepping in debugging and single-step browsing modes has
 changed slightly: now left braces at the start of the body are
 stepped over for <code>if</code> statements as well as for <code>for</code> and
 <code>while</code> statements.  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14814">PR#14814</a>.)
 </p>
 </li>
 <li> <p><code>library()</code> no longer warns about a conflict with a
 function from <code>package:base</code> if the function has the same
 code as the base one but with a different environment.  (An
 example is <code>Matrix::det()</code>.)
 </p>
 </li>
 <li><p> When deparsing very large language objects,
 <code>as.character()</code> now inserts newlines after each line of
 approximately 500 bytes, rather than truncating to the first
 line.
 </p>
 </li>
 <li><p> New function <code>rWishart()</code> generates Wishart-distributed
 random matrices.
 </p>
 </li>
 <li><p> Packages may now specify actions to be taken when the
 package is loaded (<code>setLoadActions()</code>).
 </p>
 </li>
 <li> <p><code>options(max.print = Inf)</code> and similar now give an
 error (instead of warnings later).
 </p>
 </li>
 <li><p> The <code>"difftime"</code> replacement method of <code>units</code>
 tries harder to preserve other attributes of the argument.  (Wish
 of <a href="https://bugs.R-project.org/show_bug.cgi?id=14839">PR#14839</a>.)
 </p>
 </li>
 <li> <p><code>poly(raw = TRUE)</code> no longer requires more unique
 points than the degree.  (Requested by John Fox.)
 </p>
 </li></ul>


 <h4>PACKAGE parallel</h4>


 <ul>
 <li><p> There is a new function <code>mcmapply()</code>, a parallel version
 of <code>mapply()</code>, and a wrapper <code>mcMap()</code>, a parallel
 version of <code>Map()</code>.
 </p>
 </li>
 <li><p> A default cluster can be registered by the new function
 <code>setDefaultCluster()</code>: this will be used by default in
 functions such as <code>parLapply()</code>.
 </p>
 </li>
 <li> <p><code>clusterMap()</code> has a new argument <code>.scheduling</code> to
 allow the use of load-balancing.
 </p>
 </li>
 <li><p> There are new load-balancing functions <code>parLapplyLB()</code>
 and <code>parSapplyLB()</code>.
 </p>
 </li>
 <li> <p><code>makePSOCKCluster()</code> has a new option <code>useXDR =
 	FALSE</code> which can be used to avoid byte-shuffling for
 serialization when all the nodes are known to be little-endian (or
 all big-endian).
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li><p> Non-ASCII vignettes without a declared encoding are no
 longer accepted.
 </p>
 </li>
 <li><p> C/C++ code in packages is now compiled with <code>-NDEBUG</code>
 to mitigate against the C/C++ function <code>assert</code> being called
 in production use.  Developers can turn this off during package
 development with <code>PKG_CPPFLAGS = -UNDEBUG</code>.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> has a new option <span class="option">--dsym</span>
 which on Mac OS X (Darwin) dumps the symbols alongside the
 &lsquo;<span class="file">.so</span>&rsquo; file: this is helpful when debugging with
 <code>valgrind</code> (and especially when installing packages into
 &lsquo;<span class="file">R.framework</span>&rsquo;).  [This can also be enabled by setting the
 undocumented environment variable <span class="env">PKG_MAKE_DSYM</span>, since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 2.12.0.]
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> will test loading under all installed
 sub-architectures even for packages without compiled code, unless
 the flag <span class="option">--no-multiarch</span> is used.  (Pure <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> packages can
 do things which are architecture-dependent: in the case which
 prompted this, looking for an icon in a Windows <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> executable.)
 </p>
 </li>
 <li><p> There is a new option <code>install.packages(type = "both")</code>
 which tries source packages if binary packages are not available,
 on those platforms where the latter is the default.
 </p>
 </li>
 <li><p> The meaning of <code>install.packages(dependencies = TRUE)</code>
 has changed: it now means to install the essential dependencies of
 the named packages plus the <span class="samp">Suggests</span>, but only the
 essential dependencies of dependencies.  To get the previous
 behaviour, specify <code>dependencies</code> as a character vector.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL --merge-multiarch</code> is now supported
 on OS X and other Unix-alikes using multiple sub-architectures.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL --libs-only</code> now by default does a
 test load on Unix-alikes as well as on Windows: suppress with
 <span class="option">--no-test-load</span>.
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li> <p><code>R CMD check</code> now gives a warning rather than a
 note if it finds inefficiently compressed datasets.  With
 <code>bzip2</code> and <code>xz</code> compression having been
 available since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.10.0, it only exceptionally makes sense to
 not use them.
 </p>
 <p>The environment variable <span class="env">_R_CHECK_COMPACT_DATA2_</span> is
 no longer consulted: the check is always done if
 <span class="env">_R_CHECK_COMPACT_DATA_</span> has a true value (its default).
 </p>
 </li>
 <li><p> Where multiple sub-architectures are to be tested,
 <code>R CMD check</code> now runs the examples and tests for all the
 sub-architectures even if one fails.
 </p>
 </li>
 <li> <p><code>R CMD check</code> can optionally report timings on
 various parts of the check: this is controlled by environment
 variable <span class="env">_R_CHECK_TIMINGS_</span> documented in &lsquo;Writing R
 Extensions&rsquo;.  Timings (in the style of <code>R CMD BATCH</code>) are
 given at the foot of the output files from running each test and
 the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> code in each vignette.
 </p>
 </li>
 <li><p> There are new options for more rigorous testing by
 <code>R CMD check</code> selected by environment variables &ndash; see
 the &lsquo;Writing R Extensions&rsquo; manual.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now warns (rather than notes) on
 undeclared use of other packages in examples and tests:
 increasingly people are using the metadata in the
 &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file to compute information about packages,
 for example reverse dependencies.
 </p>
 </li>
 <li><p> The defaults for some of the options in <code>R CMD
 	 check</code> (described in the &lsquo;R Internals&rsquo; manual) have
 changed: checks for unsafe and <code>.Internal()</code> calls and
 for partial matching of arguments in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> function calls are now
 done by default.
 </p>
 </li>
 <li> <p><code>R CMD check</code> has more comprehensive facilities for
 checking compiled code and so gives fewer reports on entry points
 linked into &lsquo;<span class="file">.so</span>&rsquo;/&lsquo;<span class="file">.dll</span>&rsquo; files from libraries
 (including C++ and Fortran runtimes).
 </p>
 <p>Checking compiled code is now done on FreeBSD (as well as the
 existing supported platforms of Linux, Mac OS X, Solaris and
 Windows).
 </p>
 </li>
 <li> <p><code>R CMD build</code> has more options for
 <span class="option">--compact-vignettes</span>: see <code>R CMD build --help</code>.
 </p>
 </li>
 <li> <p><code>R CMD build</code> has a new option <span class="option">--md5</span> to
 add an &lsquo;<span class="file">MD5</span>&rsquo; file (as done by CRAN): this is used by <code>R
 	 CMD INSTALL</code> to check the integrity of the distribution.
 </p>
 <p>If this option is not specified, any existing (and probably
 stale) &lsquo;<span class="file">MD5</span>&rsquo; file is removed.
 </p>
 </li></ul>


 <h4>DEPRECATED AND DEFUNCT</h4>


 <ul>
 <li> <p><code>R CMD Rd2dvi</code> is now defunct: use <code>R CMD Rd2pdf</code>.
 </p>
 </li>
 <li><p> Options such <span class="option">--max-nsize</span>, <span class="option">--max-vsize</span> and
 the function <code>mem.limits()</code> are now defunct.  (Options
 <span class="option">--min-nsize</span> and <span class="option">--min-vsize</span> remain available.)
 </p>
 </li>
 <li><p> Use of <code>library.dynam()</code> without specifying all the
 first three arguments is now disallowed.
 </p>
 <p>Use of an argument <code>chname</code> in <code>library.dynam()</code>
 including the extension &lsquo;<span class="file">.so</span>&rsquo; or &lsquo;<span class="file">.dll</span>&rsquo; (which was never
 allowed according to the help page) is defunct.  This also applies
 to <code>library.dynam.unload()</code> and to <code>useDynLib</code>
 directives in &lsquo;<span class="file">NAMESPACE</span>&rsquo; files.
 </p>
 </li>
 <li><p> The internal functions <code>.readRDS()</code> and
 <code>.saveRDS()</code> are now defunct.
 </p>
 </li>
 <li><p> The off-line <code>help()</code> types <span class="samp">"postscript"</span> and
 <span class="samp">"ps"</span> are defunct.
 </p>
 </li>
 <li> <p><code>Sys.putenv()</code>, replaced and deprecated in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.5.0, is
 finally removed.
 </p>
 </li>
 <li><p> Some functions/objects which have been defunct for five or
 more years have been removed completely.  These include
 <code>.Alias()</code>, <code>La.chol()</code>, <code>La.chol2inv()</code>,
 <code>La.eigen()</code>, <code>Machine()</code>, <code>Platform()</code>,
 <code>Version</code>, <code>codes()</code>, <code>delay()</code>,
 <code>format.char()</code>, <code>getenv()</code>, <code>httpclient()</code>,
 <code>loadURL()</code>, <code>machine()</code>, <code>parse.dcf()</code>,
 <code>printNoClass()</code>, <code>provide()</code>, <code>read.table.url()</code>,
 <code>restart()</code>, <code>scan.url()</code>, <code>symbol.C()</code>,
 <code>symbol.For()</code> and <code>unix()</code>.
 </p>
 </li>
 <li><p> The <code>ENCODING</code> argument to <code>.C()</code> is deprecated.
 It was intended to smooth the transition to multi-byte character
 strings, but can be replaced by the use of <code>iconv()</code> in the
 rare cases where it is still needed.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> Building with a positive value of
 <span class="option">--with-valgrind-instrumentation</span> now also instruments logical,
 complex and raw vectors.
 </p>
 </li></ul>


 <h4>C-LEVEL FACILITIES</h4>


 <ul>
 <li><p> Passing <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> objects other than atomic vectors, functions,
 lists and environments to <code>.C()</code> is now deprecated and will
 give a warning.  Most cases (especially <code>NULL</code>) are actually
 coding errors.  <code>NULL</code> will be disallowed in future.
 </p>
 <p><code>.C()</code> now passes a pairlist as a <code>SEXP</code> to the compiled
 code.  This is as was documented, but pairlists were in reality
 handled differently as a legacy from the early days of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 </p>
 </li>
 <li> <p><code>call_R</code> and <code>call_S</code> are deprecated.  They still
 exist in the headers and as entry points, but are no longer
 documented and should not be used for new code.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>str(x, width)</code> now obeys its <code>width</code>
 argument also for function headers and other objects <code>x</code>
 where <code>deparse()</code> is applied.
 </p>
 </li>
 <li><p> The convention for <code>x %/% 0L</code> for integer-mode <code>x</code>
 has been changed from <code>0L</code> to <code>NA_integer_</code>.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14754">PR#14754</a>)
 </p>
 </li>
 <li><p> The <code>exportMethods</code> directive in a &lsquo;<span class="file">NAMESPACE</span>&rsquo;
 file now exports S4 generics as necessary, as the extensions
 manual said it does.  The manual has also been updated to be a
 little more informative on this point.
 </p>
 <p>It is now required that there is an S4 generic (imported or
 created in the package) when methods are to be exported.
 </p>
 </li>
 <li><p> Reference methods cannot safely use non-exported entries
 in the namespace.  We now do not do so, and warn in the documentation.
 </p>
 </li>
 <li><p> The namespace import code was warning when identical
 S4 generic functions were imported more than once, but should
 not (reported by Brian Ripley, then Martin Morgan).
 </p>
 </li>
 <li> <p><code>merge()</code> is no longer allowed (in some ways) to create
 a data frame with duplicate column names (which confused
 <a href="https://bugs.R-project.org/show_bug.cgi?id=14786">PR#14786</a>).
 </p>
 </li>
 <li><p> Fixes for rendering raster images on X11 and Windows devices
 when the x-axis or y-axis scale is reversed.
 </p>
 </li>
 <li> <p><code>getAnywhere()</code> found S3 methods as seen from the
 <span class="pkg">utils</span> namespace and not from the environment from which it
 was called.
 </p>
 </li>
 <li> <p><code>selectMethod(f, sig)</code> would not return inherited group
 methods when caching was off (as it is by default).
 </p>
 </li>
 <li> <p><code>dev.copy2pdf(out.type = "cairo")</code> gave an error.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14827">PR#14827</a>)
 </p>
 </li>
 <li><p> Virtual classes (e.g., class unions) had a <code>NULL</code>
 prototype even if that was not a legal subclass.  See
 <code>?setClassUnion</code>.
 </p>
 </li>
 <li><p> The C prototypes for <code>zdotc</code> and <code>zdotu</code> in
 &lsquo;<span class="file">R_ext/BLAS.h</span>&rsquo; have been changed to the more modern style
 rather than that used by <code>f2c</code>.  (Patch by Berwin Turlach.)
 </p>
 </li>
 <li> <p><code>isGeneric()</code> produced an error for primitives that
 can not have methods.
 </p>
 </li>
 <li> <p><code>.C()</code> or <code>.Fortran()</code> had a lack-of-protection
 error if the registration information resulted in an argument
 being coerced to another type.
 </p>
 </li>
 <li> <p><code>boxplot(x=x, at=at)</code> with non finite elements in <code>x</code>
 and non integer <code>at</code> could not generate a warning but failed.
 </p>
 </li>
 <li> <p><code>heatmap(x, symm=TRUE, RowSideColors=*)</code> no longer
 draws the colors in reversed order.
 </p>
 </li>
 <li> <p><code>predict(&lt;ar&gt;)</code> was incorrect in the multivariate case,
 for p &gt;= 2.
 </p>
 </li>
 <li> <p><code>print(x, max=m)</code> is now consistent when <code>x</code> is a
 <code>"Date"</code>; also the &ldquo;reached ... max.print ..&rdquo; messages
 are now consistently using single brackets.
 </p>
 </li>
 <li><p> Closed the <span class="samp">&lt;li&gt;</span> tag in pages generated by <code>
       Rd2HTML()</code>. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14841">PR#14841</a>)
 </p>
 </li>
 <li><p> Axis tick marks could go out of range when a log scale was
 used.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14833">PR#14833</a>)
 </p>
 </li>
 <li><p> Signature objects in methods were not allocated as S4
 objects (caused a problem with <code>trace()</code> reported by Martin Morgan).
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.14.2</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li><p> The internal <code>untar()</code> (as used by default by
 <code>R CMD INSTALL</code>) now knows about some <code>pax</code>
 headers which <code>bsdtar</code> (e.g., the default <code>tar</code> for
 Mac OS &gt;= 10.6) can incorrectly include in <code>tar</code> files,
 and will skip them with a warning.
 </p>
 </li>
 <li><p> PCRE has been upgraded to version 8.21: as well as bug fixes
 and greater Perl compatibility, this adds a JIT pattern compiler,
 about which PCRE's news says
 &lsquo;large performance benefits can be had in many situations&rsquo;.
 This is supported on most but not all <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> platforms.
 </p>
 </li>
 <li><p> Function <code>compactPDF()</code> in package <span class="pkg">tools</span> now
 takes the default for argument <code>gs_quality</code> from environment
 variable <span class="env">GS_QUALITY</span>: there is a new value <code>"none"</code>, the
 ultimate default, which prevents GhostScript being used in
 preference to <code>qpdf</code> just because environment variable
 <span class="env">R_GSCMD</span> is set.  If <span class="env">R_GSCMD</span> is unset or set to
 <code>""</code>, the function will try to find a suitable GhostScript
 executable.
 </p>
 </li>
 <li><p> The included version of zlib has been updated to 1.2.6.
 </p>
 </li>
 <li><p> For consistency with the <code>logLik()</code> method,
 <code>nobs()</code> for <code>"nls"</code> files now excludes observations
 with zero weight.  (Reported by Berwin Turlach.)
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li> <p><code>R CMD check</code> now reports by default on licenses not
 according to the description in &lsquo;Writing R Extensions&rsquo;.
 </p>
 </li>
 <li> <p><code>R CMD check</code> has a new option <span class="option">--as-cran</span> to
 turn on most of the customizations that CRAN uses for its incoming
 checks.
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li> <p><code>R CMD INSTALL</code> will now no longer install certain
 file types from &lsquo;<span class="file">inst/doc</span>&rsquo;: these are almost certainly
 mistakes and for some packages are wasting a lot of space.  These
 are &lsquo;<span class="file">Makefile</span>&rsquo;, files generated by running LaTeX, and unless
 the package uses a &lsquo;<span class="file">vignettes</span>&rsquo; directory, PostScript and
 image bitmap files.
 </p>
 <p>Note that only PDF vignettes have ever been supported: some of
 these files come from DVI/PS output from the Sweave defaults prior
 to <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.0.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li><p> R configured with <span class="option">--disable-openmp</span> would mistakenly
 set <code>HAVE_OPENMP</code> (internal) and <code>SUPPORT_OPENMP</code> (in
 &lsquo;<span class="file">Rconfig.h</span>&rsquo;) even though no OpenMP flags were populated.
 </p>
 </li>
 <li><p> The <code>getS3method()</code> implementation had an old
 computation to find an S4 default method.
 </p>
 </li>
 <li> <p><code>readLines()</code> could overflow a buffer if the last line
 of the file was not terminated.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14766">PR#14766</a>)
 </p>
 </li>
 <li> <p><code>R CMD check</code> could miss undocumented S4 objects in
 packages which used S4 classes but did not <span class="samp">Depends: methods</span>
 in their &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file.
 </p>
 </li>
 <li><p> The HTML Help Search page had malformed links.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14769">PR#14769</a>)
 </p>
 </li>
 <li><p> A couple of instances of lack of protection of <code>SEXP</code>s
 have been squashed. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14772">PR#14772</a>, <a href="https://bugs.R-project.org/show_bug.cgi?id=14773">PR#14773</a>)
 </p>
 </li>
 <li> <p><code>image(x, useRaster=TRUE)</code> misbehaved on single-column
 <code>x</code>. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14774">PR#14774</a>)
 </p>
 </li>
 <li><p> Negative values for <code>options("max.print")</code> or the <code>max</code>
 argument to <code>print.default()</code> caused crashes.  Now the former
 are ignored and the latter trigger an error.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14779">PR#14779</a>)
 </p>
 </li>
 <li><p> The text of a function body containing more than 4096
 bytes was not properly saved by the parser when entered
 at the console.
 </p>
 </li>
 <li><p> Forgetting the <code>#endif</code> tag in an Rd file could cause
 the parser to go into a loop.  (Reported by Hans-Jorg Bibiko.)
 </p>
 </li>
 <li> <p><code>str(*, ....., strict.width="cut")</code> now also obeys
 <code>list.len = n</code>.  (Reported by Sören Vogel.)
 </p>
 </li>
 <li><p> Printing of arrays did not have enough protection (C level),
 e.g., in the context of <code>capture.output()</code>.  (Reported by
 Hervé Pagès and Martin Morgan.)
 </p>
 </li>
 <li> <p><code>pdf(file = NULL)</code> would produce a spurious file named
 &lsquo;<span class="file">NA</span>&rsquo;.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14808">PR#14808</a>)
 </p>
 </li>
 <li> <p><code>list2env()</code> did not check the type of its <code>envir</code>
 argument. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14807">PR#14807</a>)
 </p>
 </li>
 <li> <p><code>svg()</code> could segfault if called with a non-existent
 file path.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14790">PR#14790</a>)
 </p>
 </li>
 <li> <p><code>make install</code> can install to a path containing
 <span class="samp">+</span> characters. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14798">PR#14798</a>)
 </p>
 </li>
 <li><p> The <code>edit()</code> function did not respect the
 <code>options("keep.source")</code> setting. (Reported by Cleridy
 Lennert.)
 </p>
 </li>
 <li> <p><code>predict.lm(*, type="terms", terms=*, se.fit=TRUE)</code> did
 not work.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14817">PR#14817</a>)
 </p>
 </li>
 <li><p> There is a partial workaround for errors in the TRE
 regular-expressions engine with named classes and repeat counts of
 at least 2 in a MBCS locale (<a href="https://bugs.R-project.org/show_bug.cgi?id=14408">PR#14408</a>): these are avoided when
 TRE is in 8-bit mode (e.g., for <code>useBytes = TRUE</code> and when all
 the data are ASCII).
 </p>
 </li>
 <li><p> The C function <code>R_ReplDLLdo1()</code> did not call top-level
 handlers.
 </p>
 </li>
 <li><p> The Quartz device was unable to detect window sessions on
 Mac OS X 10.7 (Lion) and higher and thus it was not used as the
 default device on the console.  Since Lion any application can use
 window sessions, so Quartz will now be the default device if the
 user's window session is active and <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> is not run via
 <code>ssh</code> which is at least close to the behavior in prior OS
 X versions.
 </p>
 </li>
 <li> <p><code>mclapply()</code> would fail in code assembling the translated
 error message if some (but not all) cores encountered an error.
 </p>
 </li>
 <li> <p><code>format.POSIXlt(x)</code> raised an arithmetic exception when
 <code>x</code> was an invalid object of class <code>"POSIXlt"</code> and parts
 were empty.
 </p>
 </li>
 <li> <p><code>installed.packages()</code> has some more protection against
 package installs going on in parallel.
 </p>
 </li>
 <li> <p><code>.Primitive()</code> could be mis-used to call
 <code>.Internal()</code> entry points.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.14.1</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>parallel::detectCores()</code> is now able to find the
 number of physical cores (rather than CPUs) on Sparc Solaris.
 </p>
 <p>It can also do so on most versions of Windows; however the default
 remains <code>detectCores(logical = TRUE)</code> on that platform.
 </p>
 </li>
 <li><p> Reference classes now keep a record of which fields are
 locked.  <code>$lock()</code> with no arguments returns the names
 of the locked fields.
 </p>
 </li>
 <li> <p><code>HoltWinters()</code> reports a warning rather than an error
 for some optimization failures (where the answer might be a
 reasonable one).
 </p>
 </li>
 <li> <p><code>tools::dependsOnPkg()</code> now accepts the shorthand
 <code>dependencies = "all"</code>.
 </p>
 </li>
 <li> <p><code>parallel::clusterExport()</code> now allows specification of
 an environment from which to export.
 </p>
 </li>
 <li><p> The <code>quartz()</code> device now does tilde expansion on its
 <code>file</code> argument.
 </p>
 </li>
 <li> <p><code>tempfile()</code> on a Unix-alike now takes the process ID
 into account.  This is needed with <a href="https://CRAN.R-project.org/package=multicore"><span class="pkg">multicore</span></a> (and as part of
 <span class="pkg">parallel</span>) because the parent and all the children share a
 session temporary directory, and they can share the C random
 number stream used to produce the unique part.  Further, two
 children can call <code>tempfile()</code> simultaneously.
 </p>
 </li>
 <li><p> Option <code>print</code> in Sweave's <code>RweaveLatex()</code> driver
 now emulates auto-printing rather than printing (which can differ
 for an S4 object by calling <code>show()</code> rather than
 <code>print()</code>).
 </p>
 </li>
 <li> <p><code>filled.contour()</code> now accepts infinite values:
 previously it might have generated invalid graphics files
 (e.g., containing NaN values).
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> On 64-bit Linux systems, <code>configure</code> now only sets
 <span class="samp">LIBnn</span> to <code>lib64</code> if &lsquo;<span class="file">/usr/lib64</span>&rsquo; exists.  This
 may obviate setting <span class="samp">LIBnn</span> explicitly on Debian-derived
 systems.
 </p>
 <p>It is still necessary to set <span class="samp">LIBnn = lib</span> (or
 <span class="samp">lib32</span>) for 32-bit builds of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> on a 64-bit OS on those Linux
 distributions capable for supporting that concept.
 </p>
 </li>
 <li> <p><code>configure</code> looks for &lsquo;<span class="file">inconsolata.sty</span>&rsquo;, and if
 not found adjusts the default <span class="env">R_RD4PDF</span> to not use it (with a
 warning, since it is needed for high-quality rendering of manuals).
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li> <p><code>R CMD INSTALL</code> will now do a test load for all
 sub-architectures for which code was compiled (rather than just
 the primary sub-architecture).
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li><p> When checking examples under more than one sub-architecture,
 <code>R CMD check</code> now uses a separate directory
 &lsquo;<span class="file">examples_arch</span>&rsquo; for each sub-architecture, and leaves the
 output in file &lsquo;<span class="file">pkgname-Ex_arch.Rout</span>&rsquo;.  Some packages
 expect their examples to be run in a clean directory ....
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>stack()</code> now gives an error if no vector column is
 selected, rather than returning a 1-column data frame (contrary to
 its documentation).
 </p>
 </li>
 <li> <p><code>summary.mlm()</code> did not handle objects where the
 formula had been specified by an expression.  (Reported by
 Helios de Rosario Martinez).
 </p>
 </li>
 <li> <p><code>tools::deparseLatex(dropBraces=TRUE)</code> could drop
 text as well as braces.
 </p>
 </li>
 <li> <p><code>colormodel = "grey"</code> (new in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.14.0))
 did not always work in <code>postscript()</code> and <code>pdf()</code>.
 </p>
 </li>
 <li> <p><code>file.append()</code> could return <code>TRUE</code> for failures.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14727">PR#14727</a>)
 </p>
 </li>
 <li> <p><code>gzcon()</code> connections are no longer subject to garbage
 collection: it was possible for this to happen when unintended
 (e.g., when calling <code>load()</code>).
 </p>
 </li>
 <li> <p><code>nobs()</code> does not count zero-weight observations for
 <code>glm()</code> fits, for consistency with <code>lm()</code>.  This affects
 the <code>BIC()</code> values reported for such <code>glm()</code> fits.
 (Spotted by Bill Dunlap.)
 </p>
 </li>
 <li> <p><code>options(warn = 0)</code> failed to end a (C-level) context
 with more than 50 accumulated warnings.  (Spotted by Jeffrey
 Horner.)
 </p>
 </li>
 <li><p> The internal <code>plot.default()</code> code did not do sanity
 checks on a <code>cex</code> argument, so invalid input could cause
 problems.  (Reported by Ben Bolker.)
 </p>
 </li>
 <li> <p><code>anyDuplicated(&lt;array&gt;, MARGIN=0)</code> no longer fails.
 (Reported by Hervé Pagès.)
 </p>
 </li>
 <li> <p><code>read.dcf()</code> removes trailing blanks: unfortunately
 on some platforms this included <code>\xa0</code> (non-breaking space)
 which is the trailing byte of a UTF-8 character.  It now only
 considers ASCII space and tab to be &lsquo;blank&rsquo;.
 </p>
 </li>
 <li><p> There was a sign error in part of the calculations for the
 variance returned by <code>KalmanSmooth()</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14738">PR#14738</a>)
 </p>
 </li>
 <li> <p><code>pbinom(10, 1e6, 0.01, log.p = TRUE)</code> was <code>NaN</code>
 thanks to the buggy fix to <a href="https://bugs.R-project.org/show_bug.cgi?id=14320">PR#14320</a> in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.11.0.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14739">PR#14739</a>)
 </p>
 </li>
 <li> <p><code>RweaveLatex()</code> now emulates auto-printing rather than
 printing, by calling <code>methods::show()</code> when auto-printing
 would.
 </p>
 </li>
 <li> <p><code>duplicated()</code> ignored <code>fromLast</code> for a
 one-column data frame.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14742">PR#14742</a>)
 </p>
 </li>
 <li> <p><code>source()</code> and related functions did not put the correct
 timestamp on the source references; <code>srcfilecopy()</code> has gained
 a new argument <code>timestamp</code> to support this fix. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14750">PR#14750</a>)
 </p>
 </li>
 <li> <p><code>srcfilecopy()</code> has gained a new argument <code>isFile</code>
 and now records the working directory, to allow debuggers to find
 the original source file.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14826">PR#14826</a>)
 </p>
 </li>
 <li><p> LaTeX conversion of Rd files did not correctly handle
 preformatted backslashes.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14751">PR#14751</a>)
 </p>
 </li>
 <li><p> HTML conversion of Rd files did not handle markup within
 tabular cells properly.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14708">PR#14708</a>)
 </p>
 </li>
 <li> <p><code>source()</code> on an empty file with <code>keep.source =
 	TRUE</code> tried to read from <code>stdin()</code>, in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.14.0 only.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14753">PR#14753</a>)
 </p>
 </li>
 <li><p> The code to check Rd files in packages would abort if
 duplicate description sections were present.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.14.0</h3>


 <h4>SIGNIFICANT USER-VISIBLE CHANGES</h4>


 <ul>
 <li><p> All packages must have a namespace, and one is created on
 installation if not supplied in the sources.  This means that any
 package without a namespace must be re-installed under this
 version of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> (but previously-installed data-only packages without
 <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> code can still be used).
 </p>
 </li>
 <li><p> The <code>yLineBias</code> of the <code>X11()</code> and
 <code>windows()</code> families of devices has been changed from 0.1 to
 0.2: this changes slightly the vertical positioning of text in the
 margins (including axis annotations).  This is mainly for
 consistency with other devices such as <code>quartz()</code> and
 <code>pdf()</code>.  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14538">PR#14538</a>.)
 </p>
 <p>There is a new graphics parameter <code>"ylbias"</code> which
 allows the y-line bias of the graphics device to be tweaked,
 including to reproduce output from earlier versions of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 </p>
 </li>
 <li><p> Labeling of the p-values in various anova tables has been
 rationalized to be either <code>"Pr(&gt;F)"</code> or <code>"Pr(&gt;Chi)"</code>
 (i.e., the <code>"Pr(F)"</code>, <code>"Pr(Chi)"</code> and <code>"P(&gt;|Chi|)"</code>
 variants have been eliminated).  Code which extracts the p value
 <em>via</em> indexing by name may need adjustment.
 </p>
 </li>
 <li> <p><code>::</code> can now be used for datasets made available for
 lazy-loading in packages with namespaces (which makes it consistent
 with its use for data-only packages without namespaces in earlier
 versions of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>).
 </p>
 </li>
 <li><p> There is a new package <span class="pkg">parallel</span>.
 </p>
 <p>It incorporates (slightly revised) copies of packages
 <a href="https://CRAN.R-project.org/package=multicore"><span class="pkg">multicore</span></a> and <a href="https://CRAN.R-project.org/package=snow"><span class="pkg">snow</span></a> (excluding MPI, PVM and NWS
 clusters).  Code written to use the higher-level API functions in
 those packages should work unchanged (apart from changing any
 references to their namespaces to a reference to <span class="pkg">parallel</span>,
 and links explicitly to <a href="https://CRAN.R-project.org/package=multicore"><span class="pkg">multicore</span></a> or <a href="https://CRAN.R-project.org/package=snow"><span class="pkg">snow</span></a> on help
 pages).
 </p>
 <p>It also contains support for multiple RNG streams following
 L'Ecuyer <em>et al</em> (2002), with support for both
 <code>mclapply</code> and <a href="https://CRAN.R-project.org/package=snow"><span class="pkg">snow</span></a> clusters.  This replaces functions
 like <code>clusterSetupRNG()</code> from <a href="https://CRAN.R-project.org/package=snow"><span class="pkg">snow</span></a> (which are not in
 <span class="pkg">parallel</span>).
 </p>
 <p>The version released for <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.14.0 contains base functionality:
 higher-level convenience functions are planned (and some are
 already available in the &lsquo;R-devel&rsquo; version of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>).
 </p>
 </li>
 <li><p> Building PDF manuals (for <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> itself or packages,
 e.g. <em>via</em> <code>R CMD check</code>) by default requires the
 LaTeX package <span class="samp">inconsolata</span>: see the section on
 &lsquo;Making the manuals&rsquo; in the &lsquo;R Installation and
 Administration Manual&rsquo;.
 </p>
 </li>
 <li> <p><code>axTicks(*, log=TRUE)</code> has changed in some cases to
 satisfy the documented behavior and be consistent.
 </p>
 </li></ul>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>txtProgressBar()</code> can write to an open connection
 instead of the console.
 </p>
 </li>
 <li><p> Non-portable package names ending in &lsquo;<span class="file">.</span>&rsquo; are no longer
 allowed.  Nor are single-character package names (<span class="pkg">R</span> was
 already disallowed).
 </p>
 </li>
 <li> <p><code>regexpr()</code> and <code>gregexpr()</code> with <code>perl =
 	TRUE</code> allows Python-style named captures.  (Wish and
 contribution of <a href="https://bugs.R-project.org/show_bug.cgi?id=14518">PR#14518</a>.)
 </p>
 </li>
 <li><p> The placement of &lsquo;plotmath&rsquo; text in the margins of
 plots done by base graphics now makes the same vertical adjustment
 as ordinary text, so using ordinary and plotmath text on the same
 margin line will seem better aligned (but not exactly aligned,
 since ordinary text has descenders below the baseline and plotmath
 places them on the baseline).  (Related to <a href="https://bugs.R-project.org/show_bug.cgi?id=14537">PR#14537</a>.)
 </p>
 </li>
 <li> <p><code>sunflowerplot()</code> now has a formula interface.
 (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14541">PR#14541</a>.)
 </p>
 </li>
 <li> <p><code>iconv()</code> has a new argument <code>toRaw</code> to handle
 encodings such as <code>UTF-16</code> with embedded nuls (as was
 possible before the <code>CHARSXP</code> cache was introduced).
 </p>
 <p>It will also accept as input the type of list generated with
 <code>toRaw = TRUE</code>.
 </p>
 </li>
 <li><p> Garbage-collecting an unused input text connection no longer
 gives a warning (since it &lsquo;connects&rsquo; to nothing outside
 <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>).
 </p>
 </li>
 <li> <p><code>read.table()</code> and <code>scan()</code> have gained a
 <code>text</code> argument, to allow reading data from a (possibly
 literal) character string.
 </p>
 </li>
 <li> <p><code>optim(*, method = .)</code> now allows <code>method = "Brent"</code>
 as an interface to <code>optimize()</code>, for use in cases such as
 <code>mle()</code> where <code>optim()</code> is used internally.
 </p>
 </li>
 <li> <p><code>mosaicplot()</code> gains a <code>border</code> argument.
 (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14550">PR#14550</a>.)
 </p>
 </li>
 <li> <p><code>smooth.spline()</code> gains a <code>tol</code> argument which
 controls how different <code>x</code> values need to be to be treated as
 distinct.  The default has been changed to be more reliable for
 inputs whose range is small compared to their maximum absolute
 value.  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14452">PR#14452</a>.)
 </p>
 </li>
 <li> <p><code>gl()</code> runs faster by avoiding calling <code>factor()</code>.
 </p>
 </li>
 <li><p> The <code>print()</code> method for <code>object.size()</code> accepts
 <span class="samp">B</span> as well as <span class="samp">b</span> as an abbreviation for &lsquo;bytes&rsquo;.
 </p>
 </li>
 <li> <p><code>unlink()</code> gains a <code>force</code> argument to work like
 <code>rm -f</code> and if possible override restrictive permissions.
 </p>
 </li>
 <li> <p><code>pbirthday()</code> and <code>qbirthday()</code> now use exact
 calculations for <code>coincident = 2</code>.
 </p>
 </li>
 <li> <p><code>unzip()</code> and <code>unz()</code> connections have been
 updated with support for more recent Zip64 features (including
 large file sizes and <code>bzip2</code> compression, but not UTF-8
 file names).
 </p>
 <p><code>unzip()</code> has a new option to restore file times from those
 recorded (in an unknown timezone) in the zip file.
 </p>
 </li>
 <li> <p><code>update.packages()</code> now accepts a character vector of
 package names for the <code>oldPkgs</code> argument.  (Suggestion of
 Tal Galili.)
 </p>
 </li>
 <li><p> The special reference class fields <code>.self</code> and
 <code>.refClassDef</code> are now read-only to prevent corrupting
 the object.
 </p>
 </li>
 <li> <p><code>decompose()</code> now returns the original series as part
 of its value, so it can be used (rather than reconstructed) when
 plotting.  (Suggestion of Rob Hyndman.)
 </p>
 </li>
 <li><p> Rao's efficient score test has been implemented for
 <code>glm</code> objects.  Specifically, the <code>add1</code>, <code>drop1</code>,
 and <code>anova</code> methods now allow <code>test = "Rao"</code>.
 </p>
 </li>
 <li><p> If a saved workspace (e.g., &lsquo;<span class="file">.RData</span>&rsquo;) contains objects
 that cannot be loaded, <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> will now start with an warning message
 and an empty workspace, rather than failing to start.
 </p>
 </li>
 <li> <p><code>strptime()</code> now accepts times such as <span class="samp">24:00</span> for
 midnight at the end of the day, for although these are disallowed by
 POSIX 1003.1-2008, ISO 8601:2004 allows them.
 </p>
 </li>
 <li><p> Assignment of <code>names()</code> to S4 objects now checks for
 a corresponding <code>"names"</code> slot, and generates a warning or
 an error if that slot is not defined.  See the section on slots
 in <code>?Classes</code>.
 </p>
 </li>
 <li><p> The default methods for <code>is.finite()</code>, <code>is.infinite()</code>
 and <code>is.nan()</code> now signal an error if their argument is not
 an atomic vector.
 </p>
 </li>
 <li><p> The formula method for <code>plot()</code> no longer places
 package <span class="pkg">stats</span> on the search path (it loads the namespace
 instead).
 </p>
 </li>
 <li><p> There now is a genuine <code>"function"</code> method for
 <code>plot()</code> rather than the generic dispatching internally to
 <code>graphics::plot.function()</code>.  It is now exported, so can be
 called directly as <code>plot.function()</code>.
 </p>
 </li>
 <li><p> The one-sided <code>ks.test()</code> allows <code>exact = TRUE</code> to
 be specified in the presence of ties (but the approximate
 calculation remains the default: the &lsquo;exact&rsquo; computation
 makes assumptions known to be invalid in the presence of ties).
 </p>
 </li>
 <li><p> The behaviour of <code>curve(add = FALSE)</code> has changed: it
 now no longer takes the default x limits from the previous plot
 (if any): rather they default to <code>c(0, 1)</code> just as the
 <code>"function"</code> method for <code>plot()</code>.  To get the previous
 behaviour use <code>curve(add = NA)</code>, which also takes the default for
 log-scaling of the x-axis from the previous plot.
 </p>
 </li>
 <li><p> Both <code>curve()</code> and the <code>plot()</code> method for
 functions have a new argument <code>xname</code> to facilitate plots
 such as <code>sin(t)</code> <em>vs</em> <code>t</code>.
 </p>
 </li>
 <li><p> The <code>local</code> argument to <code>source()</code> can specify an
 environment as well as <code>TRUE</code> (<code>parent.env()</code>) and
 <code>FALSE</code> (<code>.GlobalEnv</code>).  It gives better error messages
 for other values, such as <code>NA</code>.
 </p>
 </li>
 <li> <p><code>vcov()</code> gains methods for classes <code>"summary.lm"</code> and
 <code>"summary.glm"</code>.
 </p>
 </li>
 <li><p> The <code>plot()</code> method for class <code>"profile.nls"</code>
 gains <code>ylab</code> and <code>lty</code> arguments, and passes
 <code>...</code> on to <code>plot.default</code>.
 </p>
 </li>
 <li><p> Character-string arguments such as the <code>mode</code> argument
 of <code>vector()</code>, <code>as.vector()</code> and <code>is.vector()</code> and
 the <code>description</code> argument of <code>file()</code> are required to
 be of length exactly one, rather than any further elements being
 silently discarded.  This helps catch incorrect usage in
 programming.
 </p>
 </li>
 <li><p> The <code>length</code> argument of <code>vector()</code> and its
 wrappers such as <code>numeric()</code> is required to be of length
 exactly one (other values are now an error rather than giving a
 warning as previously).
 </p>
 </li>
 <li> <p><code>vector(len)</code> and <code>length(x) &lt;- len</code> no longer
 accept <code>TRUE</code>/<code>FALSE</code> for <code>len</code> (not that they
 were ever documented to, but there was special-casing in the C
 code).
 </p>
 </li>
 <li><p> There is a new function <code>Sys.setFileTime()</code> to set the
 time of a file (including a directory).  See its help for exactly
 which times it sets on various OSes.
 </p>
 </li>
 <li><p> The file times reported by <code>file.info()</code> are reported
 to sub-second resolution on systems which support it.  (Currently
 the POSIX 2008 and FreeBSD/Darwin/NetBSD methods are detected.)
 </p>
 </li>
 <li><p> New function <code>getCall(m)</code> as an abstraction for
 <code>m$call</code>, enabling <code>update()</code>'s default method to apply
 more universally.  (NB: this can be masked by existing functions
 in packages.)
 </p>
 </li>
 <li> <p><code>Sys.info()</code> gains a <code>euser</code> component to report
 the &lsquo;effective&rsquo; user on OSes which have that concept.
 </p>
 </li>
 <li><p> The result returned by <code>try()</code> now contains the
 original error condition object as the <code>"condition"</code>
 attribute.
 </p>
 </li>
 <li><p> All packages with <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> code are lazy-loaded irrespective of
 the <span class="samp">LazyLoad</span> field in the &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file.  A
 warning is given if the <span class="samp">LazyLoad</span> field is overridden.
 </p>
 </li>
 <li> <p><code>Rd</code> markup has a new <span class="samp">\figure</span> tag so that
 figures can be included in help pages when converted to HTML or
 LaTeX.  There are examples on the help pages for <code>par()</code> and
 <code>points()</code>.
 </p>
 </li>
 <li><p> The built-in httpd server now allows access to files
 in the session temporary directory <code>tempdir()</code>,
 addressed as the &lsquo;<span class="file">/session</span>&rsquo; directory on the httpd
 server.
 </p>
 </li>
 <li><p> Development versions of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> are no longer referred to by the
 number under which they might be released, e.g., in the startup
 banner, <code>R --version</code> and <code>sessionUtils()</code>.  The
 correct way to refer to a development version of R is
 &lsquo;R-devel&rsquo;, preferably with the date and SVN version number.<br />
 E.g., <span class="samp">R-devel (2011-07-04 r56266)</span>
 </p>
 </li>
 <li><p> There is a new function <code>texi2pdf()</code> in package
 <span class="pkg">tools</span>, currently a convenience wrapper for
 <code>texi2dvi(pdf = TRUE)</code>.
 </p>
 </li>
 <li><p> There are two new options for typesetting PDF manuals from
 Rd files.  These are <span class="samp">beramono</span> and <span class="samp">inconsolata</span>, and
 used the named font for monospaced output. They are intended to
 be used in combination with <span class="samp">times</span>, and
 <span class="samp">times,inconsolata,hyper</span> is now the default for the
 reference manual and package manuals.  If you do not have that
 font installed, you can set <span class="env">R_RD4PF</span> to one of the other
 options: see the &lsquo;R Installation and Administration Manual&rsquo;.
 </p>
 </li>
 <li><p> Automatic printing for reference classes is now done by
 the <code>$show()</code> method.  A method is defined for class
 <span class="samp">envRefClass</span> and may be overridden for user classes (see
 the <code>?ReferenceClasses</code> example).  S4 <code>show()</code> methods
 should no longer be needed for reference classes.
 </p>
 </li>
 <li> <p><code>tools::Rdiff</code> (by default) and <code>R CMD Rdiff</code>
 now ignore differences in pointer values when comparing printed
 environments, compiled byte code, etc.
 </p>
 </li>
 <li><p> The <code>"source"</code> attribute on functions created with
 <code>keep.source=TRUE</code> has been replaced with a <code>"srcref"</code>
 attribute.  The <code>"srcref"</code> attribute references an in-memory
 copy of the source file using the <code>"srcfilecopy"</code> class or
 the new <code>"srcfilealias"</code> class.
 </p>
 <p><strong>NB:</strong> This means that functions sourced with
 <code>keep.source = TRUE</code> and saved (e.g., by <code>save()</code> or
 <code>readRDS()</code>) in earlier versions of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> will no longer show
 the original sources (including comments).
 </p>
 </li>
 <li><p> New items <span class="samp">User Manuals</span> and
 <a href="http://developer.r-project.org/TechDocs">Technical Papers</a>
 have been added to the HTML help main page.  These link to
 vignettes in the base and recommended packages and to a
 collection of papers about R issues, respectively.
 </p>
 </li>
 <li><p> Documentation and messages have been standardized to
 use &ldquo;namespace&rdquo; rather than &ldquo;name space&rdquo;.
 </p>
 </li>
 <li> <p><code>setGeneric()</code> now looks in the default packages for
 a non-generic version of a function if called from a package with a
 namespace.  (It always did for packages without a namespace.)
 </p>
 </li>
 <li><p> Setting the environment variable
 <span class="env">_R_WARN_ON_LOCKED_BINDINGS_</span> will give a warning if an
 attempt is made to change a locked binding.
 </p>
 </li>
 <li> <p><code>\SweaveInput</code> is now supported when generating
 concordances in <code>Sweave()</code>.
 </p>
 </li>
 <li> <p><code>findLineNum()</code> and <code>setBreakpoint()</code> now allow
 the environment to be specified indirectly; the latter gains
 a <code>clear</code> argument to allow it to call <code>untrace()</code>.
 </p>
 </li>
 <li><p> The body of a closure can be one of further types of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 objects, including environments and external pointers.
 </p>
 </li>
 <li><p> The <code>Rd2HTML()</code> function in package <span class="pkg">tools</span> now has
 a <code>stylesheet</code> argument, allowing pages to be displayed in
 alternate formats.
 </p>
 </li>
 <li><p> New function <code>requireNamespace()</code> analogous to
 <code>require()</code>, returning a logical value after attempting to
 load a namespace.
 </p>
 </li>
 <li><p> There is a new type of RNG, <code>"L'Ecuyer-CMRG"</code>,
 implementing L'Ecuyer (1999)'s &lsquo;combined multiple-recursive
 generator&rsquo; <span class="samp">MRG32k3a</span>.  See the comments on <code>?RNG</code>.
 </p>
 </li>
 <li> <p><code>help.search()</code> and <code>??</code> can now display
 vignettes and demos as well as help pages.  The new option
 <code>"help.search.types"</code> controls the types of documentation
 and the order of their display.
 </p>
 <p>This also applies to HTML searches, which now give results in all
 of help pages, vignettes and demos.
 </p>
 </li>
 <li> <p><code>socketConnection()</code> now has a <code>timeout</code> argument.
 It is now documented that large values (package <a href="https://CRAN.R-project.org/package=snow"><span class="pkg">snow</span></a> used a
 year) do not work on some OSes.
 </p>
 </li>
 <li><p> The initialization of the random-number generator now uses
 the process ID as well as the current time, just in case two <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 processes are launched very rapidly on a machine with
 low-resolution wall clock (some have a resolution of a second;
 modern systems have microsecond-level resolution).
 </p>
 </li>
 <li><p> New function <code>pskill()</code> in the <span class="pkg">tools</span> package to
 send a terminate signal to one or more processes, plus constants
 such as <code>SIGTERM</code> to provide a portable way to refer to
 signals (since the numeric values are OS-dependent).
 </p>
 </li>
 <li><p> New function <code>psnice()</code> in the <span class="pkg">tools</span> package to
 return or change the &lsquo;niceness&rsquo; of a process.  (Refers to
 the &lsquo;priority class&rsquo; on Windows.)
 </p>
 </li>
 <li> <p><code>list.dirs()</code> gains a <code>recursive</code> argument.
 </p>
 </li>
 <li><p> An <span class="samp">Authors@R</span> field in a package &lsquo;<span class="file">DESCRIPTION</span>&rsquo;
 file can now be used to generate <span class="samp">Author</span> and
 <span class="samp">Maintainer</span> fields if needed, and to auto-generate package
 citations.
 </p>
 </li>
 <li><p> New utility <code>getElement()</code> for accessing either a list
 component or a slot in an S4 object.
 </p>
 </li>
 <li> <p><code>stars()</code> gains a <code>col.lines</code> argument, thanks to
 Dustin Sallings. (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14657">PR#14657</a>.)
 </p>
 </li>
 <li><p> New function <code>regmatches()</code> for extracting or replacing
 matched or non-matched substrings from match data obtained by
 <code>regexpr()</code>, <code>gregexpr()</code> and <code>regexec()</code>.
 </p>
 </li>
 <li> <p><code>help(package = "pkg_name", help_type = "HTML")</code> now
 gives HTML help on the package rather than text help.  (This gives
 direct access to the HTML version of the package manual shown
 <em>via</em> <code>help.start()</code>'s &lsquo;Packages&rsquo; menu.)
 </p>
 </li>
 <li> <p><code>agrep()</code> gains a <code>fixed</code> argument to optionally
 allow approximate regular expression matching, and a <code>costs</code>
 argument to specify possibly different integer match costs for
 insertions, deletions and substitutions.
 </p>
 </li>
 <li> <p><code>read.dcf()</code> and <code>write.dcf()</code> gain a
 <code>keep.white</code> argument to indicate fields where whitespace
 should be kept as is.
 </p>
 </li>
 <li> <p><code>available.packages()</code> now works around servers that
 fail to return an error code when &lsquo;<span class="file">PACKAGES.gz</span>&rsquo; does not
 exist.  (Patch submitted by Seth Schommer.)
 </p>
 </li>
 <li> <p><code>readBin()</code> can now read more than
 <i>2^31 - 1</i> bytes in a single call (the previously
 documented limitation).
 </p>
 </li>
 <li><p> New function <code>regexec()</code> for finding the positions of
 matches as well as all substrings corresponding to parenthesized
 subexpressions of the given regular expression.
 </p>
 </li>
 <li><p> New function <code>adist()</code> in package <span class="pkg">utils</span> for
 computing &lsquo;edit&rsquo; (generalized Levenshtein) distances between
 strings.
 </p>
 </li>
 <li><p> Class <code>"raster"</code> gains an <code>is.na</code> method to avoid
 confusion from the misuse of the matrix method (such as <a href="https://bugs.R-project.org/show_bug.cgi?id=14618">PR#14618</a>).
 </p>
 </li>
 <li><p> The <code>identical()</code> function gains an
 <code>ignore.bytecode</code> argument to control comparison of
 compiled functions.
 </p>
 </li>
 <li> <p><code>pmin</code> and <code>pmax</code> now warn if an argument is
 partially recycled (wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14638">PR#14638</a>).
 </p>
 </li>
 <li><p> The default for <code>image(useRaster=)</code> is now taken from
 option <code>"preferRaster"</code>: for the small print see <code>?image</code>.
 </p>
 </li>
 <li> <p><code>str()</code> now displays reference class objects and their
 fields, rather than treating them as classical S4 classes.
 </p>
 </li>
 <li><p> New function <code>aregexec()</code> in package <span class="pkg">utils</span> for
 finding the positions of approximate string matches as well as all
 substrings corresponding to parenthesized subexpressions of the
 given regular expression.
 </p>
 </li>
 <li> <p><code>download.file()</code> has an <code>extra</code> argument to pass
 additional command-line options to the non-default methods using
 command-line utilities.
 </p>
 <p><code>cacheOK = FALSE</code> is now supported for <code>method = "curl"</code>.
 </p>
 </li>
 <li> <p><code>interaction.plot(*, type = .)</code> now also allows type
 <code>"o"</code> or <code>"c"</code>.
 </p>
 </li>
 <li> <p><code>axTicks(*, log=TRUE)</code> did sometimes give more values
 than the ticks in the corresponding <code>graphics::axis()</code>.  By
 default, it now makes use of the new (<span class="pkg">graphics</span>-package
 independent) <code>axisTicks()</code> which can make use of a new
 utility <code>.axisPars()</code>.  Further, it now returns a decreasing
 sequence (as for <code>log=FALSE</code>) when <code>usr</code> is decreasing.
 </p>
 </li>
 <li><p> Using <code>fix()</code> or <code>edit()</code> on a <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> object (except
 perhaps a matrix or data frame) writes its temporary file with
 extension &lsquo;<span class="file">.R</span>&rsquo; so editors which select their mode based on
 the extension will select a suitable mode.
 </p>
 </li></ul>


 <h4>GRAPHICS DEVICES</h4>


 <ul>
 <li><p> The <code>pdf()</code> device makes use of Flate compression: this
 is controlled by the new logical argument <code>compress</code>, and is
 enabled by default.
 </p>
 </li>
 <li><p> Devices <code>svg()</code>, <code>cairo_pdf()</code> and
 <code>cairo_ps()</code> gain a <code>family</code> argument.
 </p>
 <p>On a Unix-alike <code>X11()</code> gains a <code>family</code> argument.  This
 is one of the <code>x11.options()</code> and so can be passed as an
 argument to the <code>bmp()</code>, <code>jpeg()</code>, <code>png()</code> and
 <code>tiff()</code> devices.
 </p>
 <p>Analogous changes have been made on Windows, so all built-in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 graphics devices now have a <code>family</code> argument except
 <code>pictex()</code> (which has no means to change fonts).
 </p>
 </li>
 <li><p> The <code>bmp()</code>, <code>jpeg()</code>, <code>png()</code> and <code>tiff()</code>
 devices now make use of the <code>antialias</code> argument for
 <code>type = "quartz"</code>.
 </p>
 </li>
 <li><p> There are several new built-in font mappings for
 <code>X11(type = "Xlib")</code>: see the help on <code>X11Fonts()</code>.
 </p>
 </li>
 <li><p> There is a new type <code>X11(type = "dbcairo")</code> which
 updates the screen less frequently: see its help page.
 </p>
 </li>
 <li><p> The <code>X11()</code> device now makes use of cursors to
 distinguish its states.  The normal cursor is an arrow (rather
 than a crosshair); the crosshair is used when the locator is in
 use, and a watch cursor is shown when plotting computations are
 being done.  (These are the standard names for X11 cursors: how
 they are actually displayed depends on the window manager.)
 </p>
 </li>
 <li><p> New functions <code>dev.hold()</code> and <code>dev.flush()</code> for
 use with graphics devices with buffering.  These are used for most
 of the high-level graphics functions such as <code>boxplot()</code>, so
 that the plot is only displayed when the page is complete.
 </p>
 <p>Currently implemented for <code>windows(buffered = TRUE)</code>,
 <code>quartz()</code> and the cairographics-based <code>X11()</code> types
 with buffering (which are the default on-screen devices).
 </p>
 </li>
 <li><p> New function <code>dev.capture()</code> for capture of bitmap
 snapshots of image-based devices (a superset of the functionality
 provided by <code>grid.cap()</code> in <code>grid</code>).
 </p>
 </li>
 <li><p> The default <code>colormodel</code> for <code>pdf()</code> and
 <code>postscript()</code> is now called <code>"srgb"</code> to more accurately
 describe it.  (Instead of <code>"rgb"</code>, and in the case of
 <code>postscript()</code> it no longer switches to and from the gray
 colorspace, by default.)
 </p>
 <p>The <code>colormodel</code> for <code>postscript()</code> which does use both
 gray and sRGB colorspaces is now called <code>"srgb+gray"</code>.
 </p>
 <p>Plots which are known to use only black/white/transparent can
 advantageously use <code>colormodel = "gray"</code> (just as before, but
 there is now slightly more advantage in doing so).
 </p>
 </li>
 <li> <p><code>postscript()</code> with values <code>colormodel = "rgb"</code> and
 <code>colormodel = "rgb-nogray"</code> give the behaviour prior to <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 2.13.0 of uncalibrated RGB, which under some circumstances can be
 rendered much faster by a viewer.
 </p>
 <p><code>pdf(colormodel = "rgb")</code> gives the behaviour prior to <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 2.13.0 of uncalibrated RGB, which under some circumstances can be
 rendered faster by a viewer, and the files will be smaller (by
 about 9KB if compression is not used).
 </p>
 </li>
 <li><p> The <code>postscript()</code> device only includes the definition
 of the sRGB colorspace in the output file for the colormodels
 which use it.
 </p>
 </li>
 <li><p> The <code>postscript()</code> and <code>pdf()</code> devices now output
 greyscale raster images (and not RGB) when <code>colormodel = "gray"</code>.
 </p>
 </li>
 <li> <p><code>postscript(colormodel = "gray")</code> now accepts non-grey
 colours and uses their luminance (as <code>pdf()</code> long has).
 </p>
 </li>
 <li> <p><code>colormodel = "grey"</code> is allowed as an alternative name
 for <code>postscript()</code> and <code>pdf()</code>.
 </p>
 </li>
 <li> <p><code>pdf()</code> in the default sRGB colorspace outputs many
 fewer changes of colorspace, which may speed up rendering in some
 viewing applications.
 </p>
 </li>
 <li><p> There is a new function <code>dev.capabilities()</code> to query
 the capabilities of the current device.  The initial set of
 capabilities are support for semi-transparent colours, rendering
 and capturing raster images, the locator and for interactive
 events.
 </p>
 </li>
 <li><p> For <code>pdf()</code>, <code>maxRasters</code> is increased as needed
 so the argument is no longer used.
 </p>
 </li></ul>


 <h4>SWEAVE &amp; VIGNETTES</h4>


 <ul>
 <li><p> Options <code>keep.source = TRUE, figs.only = FALSE</code>
 are now the default.
 </p>
 </li>
 <li><p> The way the type of user-defined options is determined has
 changed.  Previously they were all regarded as logical: now the
 type is determined by the value given at first use.
 </p>
 </li>
 <li><p> The allowed values of logical options are now precisely
 those allowed for character inputs to <code>as.logical()</code>: this
 means that <span class="samp">t</span> and <span class="samp">f</span> are no longer allowed (although
 <code>T</code> and <code>F</code> still are).
 </p>
 </li>
 <li><p> The preferred location for vignette sources is now the
 directory &lsquo;<span class="file">vignettes</span>&rsquo; and not &lsquo;<span class="file">inst/doc</span>&rsquo;: <code>R CMD
       build</code> will now re-build vignettes in directory &lsquo;<span class="file">vignettes</span>&rsquo;
 and copy the &lsquo;<span class="file">.Rnw</span>&rsquo; (etc) files and the corresponding PDFs to
 &lsquo;<span class="file">inst/doc</span>&rsquo;.  Further files to be copied to &lsquo;<span class="file">inst/doc</span>&rsquo;
 can be specified <em>via</em> the file
 &lsquo;<span class="file">vignettes/.install_extras</span>&rsquo;.
 </p>
 </li>
 <li> <p><code>R CMD Sweave</code> now supports a <span class="option">--driver</span>
 option to select the Sweave driver: the default is equivalent to
 <span class="option">--driver=RweaveLatex</span>.
 </p>
 </li>
 <li> <p><code>R CMD Sweave</code> and <code>R CMD Stangle</code> support
 options <span class="option">--encoding</span> and <span class="option">--options</span>.
 </p>
 </li>
 <li><p> The <code>Rtangle()</code> driver allows <code>output = "stdout"</code>
 or <code>output = "stderr"</code> to select the output or message
 connection.  This is convenient for scripting using something like
 </p>
 <pre>    R CMD Stangle --options='output="stdout"' foo.Rnw &gt; foo2.R</pre>
 </li>
 <li><p> There is a new option <code>pdf.compress</code> controlling
 whether PDF figures are generated using Flate compression (they
 are by default).
 </p>
 </li>
 <li> <p><code>R CMD Sweave</code> now has a <span class="option">--pdf</span> option to
 produce a PDF version of the processed Sweave document.
 </p>
 </li>
 <li><p> It is no longer allowed to have two vignettes with the
 same vignette basename (e.g., &lsquo;<span class="file">vig.Rnw</span>&rsquo; and &lsquo;<span class="file">vig.Snw</span>&rsquo;).
 (Previously one vignette hid the other in the <code>vignette()</code>
 function.)
 </p>
 </li></ul>


 <h4>C-LEVEL FACILITIES</h4>


 <ul>
 <li><p> Function <code>R_tmpnam2</code> has been added to the API to allow
 a temporary filename to include a specified extension.
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li><p> Package &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file field <span class="samp">KeepSource</span>
 forces the package to be installed with <code>keep.source =
 	TRUE</code> (or <code>FALSE</code>).  (Suggestion of Greg Snow.  Note that
 as all packages are lazy-loaded, this is now only relevant at
 installation.)
 </p>
 <p>There are corresponding options <span class="option">--with-keep.source</span> and
 <span class="option">--without-keep.source</span> for <code>R CMD INSTALL</code>.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> has a new option
 <span class="option">--byte-compile</span> to byte-compile the packages during
 installation (since all packages are now lazy-loaded).  This can
 be controlled on a per-package basis by the optional field
 <span class="samp">ByteCompile</span> in the &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file.
 </p>
 </li>
 <li><p> A package <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> code but without a &lsquo;<span class="file">NAMESPACE</span>&rsquo; file will
 have a default one created at <code>R CMD build</code> or <code>R
       CMD INSTALL</code> time, so all packages will be installed with
 namespaces.  A consequence of this is that <code>.First.lib()</code>
 functions need to be copied to <code>.onLoad()</code> (usually) or
 <code>.onAttach()</code>.  For the time being, if there is an
 auto-generated &lsquo;<span class="file">NAMESPACE</span>&rsquo; file and no <code>.onLoad()</code> nor
 <code>.onAttach()</code> function is found but <code>.First.lib()</code> is,
 it will be run as the attach hook (unless the package is one of a
 list of known exceptions, when it will be run as the load hook).
 </p>
 </li>
 <li><p> A warning is given if test-loading a package changes a
 locked binding in a package other than itself.  It is likely that
 this will be disallowed in future releases.  (There are <em>pro
 tem</em> some exceptions to the warning.)
 </p>
 </li>
 <li><p> A dependency on SVN revision is allowed for <code>R</code>,
 e.g. <code>R (&gt;= r56550)</code>.  This should be used in conjunction with a
 version number, e.g. <code>R (&gt;= 2.14.0), R (&gt;= r56550)</code> to
 distinguish between R-patched and R-devel versions with the same
 SVN revision.
 </p>
 </li>
 <li> <p><code>installed.packages()</code> now hashes the names of its
 cache files to avoid very rare problems with excessively long path
 names.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14669">PR#14669</a>)
 </p>
 </li>
 <li><p> A top-level &lsquo;<span class="file">COPYING</span>&rsquo; file in a package is no longer
 installed (file names &lsquo;<span class="file">LICENSE</span>&rsquo; or &lsquo;<span class="file">LICENCE</span>&rsquo; having long
 been preferred).
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li> <p><code>R CMD check</code> now gives an error if the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> code in a
 vignette fails to run, unless this is caused by a missing package.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now unpacks tarballs in the same way
 as <code>R CMD INSTALL</code>, including making use of the
 environment variable <span class="env">R_INSTALL_TAR</span> to override the default
 behaviour.
 </p>
 </li>
 <li> <p><code>R CMD check</code> performs additional code analysis of
 package startup functions, and notifies about incorrect argument
 lists and (incorrect) calls to functions which modify the search
 path or inappropriately generate messages.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now also checks compiled code for
 symbols corresponding to functions which might terminate <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> or
 write to &lsquo;<span class="file">stdout</span>&rsquo;/&lsquo;<span class="file">stderr</span>&rsquo; instead of the console.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now uses a <code>pdf()</code> device when
 checking examples (rather than <code>postscript()</code>).
 </p>
 </li>
 <li> <p><code>R CMD check</code> now checks line-endings of makefiles
 and C/C++/Fortran sources in subdirectories of &lsquo;<span class="file">src</span>&rsquo; as well
 as in &lsquo;<span class="file">src</span>&rsquo; itself.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now reports as a NOTE what look like
 methods documented with their full names even if there is a
 namespace and they are exported.  In almost all cases they are
 intended to be used only as methods and should use the
 <span class="samp">\method</span> markup.  In the other rare cases the recommended
 form is to use a function such as <code>coefHclust</code> which would
 not get confused with a method, document that and register it in
 the &lsquo;<span class="file">NAMESPACE</span>&rsquo; file by <code>s3method(coef, hclust,
       coefHclust)</code>.
 </p>
 </li>
 <li><p> The default for the environment variable
 <span class="env">_R_CHECK_COMPACT_DATA2_</span> is now true: thus if using the newer
 forms of compression introduced in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.10.0 would be beneficial
 is now checked (by default).
 </p>
 </li>
 <li><p> Reference output for a vignette can be supplied when
 checking a package by <code>R CMD check</code>: see &lsquo;Writing R
 Extensions&rsquo;.
 </p>
 </li>
 <li> <p><code>R CMD Rd2dvi</code> allows the use of LaTeX package
 <span class="samp">inputenx</span> rather than <span class="samp">inputenc</span>: the value of the
 environment variable <span class="env">RD2DVI_INPUTENC</span> is used.  (LaTeX package
 <span class="samp">inputenx</span> is an optional install which provides greater
 coverage of the UTF-8 encoding.)
 </p>
 </li>
 <li> <p><code>Rscript</code> on a Unix-alike now accepts file names
 containing spaces (provided these are escaped or quoted in the
 shell).
 </p>
 </li>
 <li> <p><code>R CMD build</code> on a Unix-alike (only) now tries to
 preserve dates on files it copies from its input directory.  (This
 was the undocumented behaviour prior to <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.0.)
 </p>
 </li></ul>


 <h4>DEPRECATED AND DEFUNCT</h4>


 <ul>
 <li> <p><code>require()</code> no longer has a <code>save</code> argument.
 </p>
 </li>
 <li><p> The <code>gamma</code> argument to <code>hsv()</code>,
 <code>rainbow()</code>, and <code>rgb2hsv()</code> has been removed.
 </p>
 </li>
 <li><p> The <span class="option">--no-docs</span> option for
 <code>R CMD build --binary</code> is defunct: use
 <span class="option">--install-args</span> instead.
 </p>
 </li>
 <li><p> The option <span class="option">--unsafe</span> to <code>R CMD INSTALL</code> is
 defunct: use the identical option <span class="option">--no-lock</span> instead.
 </p>
 </li>
 <li><p> The entry point <code>pythag</code> formerly in &lsquo;<span class="file">Rmath.h</span>&rsquo; is
 defunct: use instead the C99 function <code>hypot</code>.
 </p>
 </li>
 <li> <p><code>R CMD build --binary</code> is formally defunct:
 <code>R CMD INSTALL --build</code> has long been the preferred
 alternative.
 </p>
 </li>
 <li> <p><code>zip.file.extract()</code> is now defunct: use
 <code>unzip()</code> or <code>unz()</code> instead.
 </p>
 </li>
 <li> <p><code>R CMD Rd2dvi</code> without the <span class="option">--pdf</span> option is
 now deprecated: only PDF output will be supported in future
 releases (since this allows the use of fonts only supported for
 PDF), and only <code>R CMD Rd2pdf</code> will be available.
 </p>
 </li>
 <li><p> Options such as <span class="option">--max-nsize</span> and the function
 <code>mem.limits()</code> are now deprecated: these limits are nowadays
 almost never used, and are reported by <code>gc()</code> when they are
 in use.
 </p>
 </li>
 <li><p> Forms like <code>binomial(link = "link")</code> for GLM families
 deprecated since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.4.0 are now defunct.
 </p>
 </li>
 <li><p> The <code>declarativeOnly</code> argument to
 <code>loadNamespace()</code> (not relevant since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.0) has been
 removed.
 </p>
 </li>
 <li><p> Use of <code>library.dynam()</code> without specifying all the
 first three arguments is deprecated.  (It is often called from a
 namespace, and the defaults are only appropriate to a package.)
 </p>
 <p>Use of <code>chname</code> in <code>library.dynam()</code> with the extension
 &lsquo;<span class="file">.so</span>&rsquo; or <code>.dll</code> (which is clearly not allowed according
 to the help page) is deprecated.  This also applies to
 <code>library.dynam.unload()</code> and <code>useDynLib</code> directives in
 &lsquo;<span class="file">NAMESPACE</span>&rsquo; files.
 </p>
 </li>
 <li><p> It is deprecated to use <code>mean(x)</code> and <code>sd(x)</code>
 directly on data frames (or also matrices, for <code>sd</code>)
 <code>x</code>, instead of simply using <code>sapply</code>.
 </p>
 <p>In the same spirit, <code>median(x)</code> now gives an error for a data
 frame <code>x</code> (it often gave nonsensical results).
 </p>
 </li>
 <li><p> The <code>keep.source</code> argument to <code>library()</code> and
 <code>require()</code> is deprecated: it was only used for packages
 installed without lazy-loading, and now all packages are
 lazy-loaded.
 </p>
 </li>
 <li><p> Using a false value for the &lsquo;<span class="file">DESCRIPTION</span>&rsquo; field
 <span class="samp">LazyLoad</span> is deprecated.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> The base and recommended packages are now byte-compiled
 (equivalent to <code>make bytecode</code> in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.x).
 </p>
 </li>
 <li><p> Configure option <span class="option">--with-system-zlib</span> now only makes
 use of the basic interface of <code>zlib</code> and not the
 C function <span class="samp">gzseek</span> which has shown erroneous behaviour in
 <code>zlib</code> 1.2.4 and 1.2.5.
 </p>
 </li>
 <li><p> The <code>zlib</code> in the R sources is now version 1.2.5.
 (This is safe even on 32-bit Linux systems because only the basic
 interface is now used.)
 </p>
 </li>
 <li><p> The &lsquo;<span class="file">.afm</span>&rsquo; files in package <span class="pkg">grDevices</span> are now
 installed as compressed files (as long done on Windows), saving ca
 2MB on the installed size.
 </p>
 </li>
 <li><p> The non-screen cairo-based devices are no longer in the X11
 module and so can be installed without X11.  (We have never seen a
 Unix-alike system with cairographics installed but not X11, but a
 user might select <span class="option">--without-x</span>.)
 </p>
 </li>
 <li><p> Configure will try to use <span class="option">-fobjc-exceptions</span> for the
 Objective-C compiler (if present) to ensure that even compilers
 that do not enable exceptions by default (such as vanilla gcc) can
 be used.  (Objective-C is currently only used on Mac OS X.)
 </p>
 </li>
 <li><p> The system call <code>times</code> is required.
 </p>
 </li>
 <li><p> The C99 functions <code>acosh</code>, <code>asinh</code>, <code>atanh</code>,
 <code>snprintf</code> and <code>vsnprintf</code> are now required.
 </p>
 </li>
 <li><p> There is no longer support for making DVI manuals <em>via</em>
 <code>make dvi</code>, <code>make install-dvi</code> and similar.  Only
 PDF manuals are supported (to allow the use of fonts which are
 only available for PDF.)
 </p>
 </li>
 <li><p> The &lsquo;<span class="file">configure</span>&rsquo; arguments used during configuration of
 <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> are included as a comment in &lsquo;<span class="file">Makeconf</span>&rsquo; for informative
 purposes on Unix-alikes in a form suitable for shell execution.
 Note that those are merely command-line arguments, they do not
 include environment variables (one more reason to use configure
 variables instead) or site configuration settings.
 </p>
 </li>
 <li><p> Framework installation now supports <code>DESTDIR</code>
 (Mac OS X only).
 </p>
 </li>
 <li><p> Java detection (<code>R CMD javareconf</code>) works around bogus
 <code>java.library.path</code> property in recent Oracle Java binaries.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li><p> The locale category <span class="samp">LC_MONETARY</span> was only being set on
 startup on Windows: that is now done on Unix-alikes where
 supported.
 </p>
 </li>
 <li><p> Reference class utilities will detect an attempt to modify
 methods or fields in a locked class definition (e.g., in a
 namespace) and generate an error.
 </p>
 </li>
 <li><p> The formula methods for <code>lines()</code>, <code>points()</code> and
 <code>text()</code> now work even if package <span class="pkg">stats</span> is not on the
 search path.
 </p>
 </li>
 <li><p> In principle, S4 classes from different packages could
 have the same name.  This has not previously worked.  Changes
 have now been installed that should allow such classes and
 permit methods to use them.  New functions <code>className()</code>
 and <code>multipleClasses()</code> are related tools for programming.
 </p>
 </li>
 <li><p> Work around an issue in Linux (a system <code>select</code> call
 resetting <code>tv</code>) which prevented internet operations from
 timing out properly.
 </p>
 </li>
 <li><p> Several stack trampling and overflow issues have been fixed
 in TRE, triggered by <code>agrep</code> and friends with long patterns.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14627">PR#14627</a>)
 </p>
 </li>
 <li><p> (&ldquo;design infelicity&rdquo;) Field assignments in reference
 classes are now consistent with slots in S4 classes:  the
 assigned value must come from the declared class (if any) for the field
 or from a subclass.
 </p>
 </li>
 <li><p> The methods objects constructed for <code>"coerce"</code> and
 <code>"coerce&lt;-"</code> were lacking some essential information in the
 <code>generic</code>, <code>defined</code> and <code>target</code> slots;
 <code>as()</code> did not handle duplicate class definitions correctly.
 </p>
 </li>
 <li><p> The parser no longer accepts the digit <code>8</code> in an octal
 character code in a string, nor does it accept unterminated
 strings in a file.  (Reported by Bill Dunlap.)
 </p>
 </li>
 <li><p> The <code>print()</code> method for class <code>"summary.aov"</code> did
 not pass on argument <code>digits</code> when <code>summary()</code> was
 called on a single object, and hence used more digits than
 documented.
 </p>
 </li>
 <li><p> The <code>X11()</code> device's cairo back-end produced incorrect
 capture snapshot images on big-endian machines.
 </p>
 </li>
 <li> <p><code>loglin()</code> gave a spurious error when argument
 <code>margin</code> consisted of a single element of length one.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14690">PR#14690</a>)
 </p>
 </li>
 <li> <p><code>loess()</code> is better protected against misuse,
 e.g. zero-length <code>span</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14691">PR#14691</a>)
 </p>
 </li>
 <li> <p><code>HoltWinters()</code> checks that the optimization
 succeeded. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14694">PR#14694</a>)
 </p>
 </li>
 <li><p> The (undocumented) inclusion of superclass objects in
 default initializing of reference classes overwrote explicit
 field arguments. The bug is fixed, the feature documented and a
 test added.
 </p>
 </li>
 <li> <p><code>round(x, -Inf)</code> now does something sensible (return
 zero rather than <code>NA</code>).
 </p>
 </li>
 <li> <p><code>signif(x, -Inf)</code> now behaves as documented
 (<code>signif(x, 1)</code>) rather than giving zero.
 </p>
 </li>
 <li><p> The <code>"table"</code> method for <code>Axis()</code> hardcoded
 <code>side = 1</code>, hence calls to <code>plot(&lt;vector&gt;, &lt;table&gt;)</code>
 labelled the wrong axis. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14699">PR#14699</a>)
 </p>
 </li>
 <li><p> Creating a connection might fail under <code>gctorture(TRUE)</code>.
 </p>
 </li>
 <li> <p><code>stack()</code> and <code>unstack()</code> converted character
 columns to factors.
 </p>
 <p><code>unstack()</code> sometimes produced incorrect results (a list or a
 vector) if the factor on which to un-split had only one level.
 </p>
 </li>
 <li><p> On some systems <code>help(".C", help_type = "pdf")</code> and
 similar generated file names that TeX was unable to handle.
 </p>
 </li>
 <li><p> Non-blocking listening socket connections continued to
 report <code>isIncomplete()</code> as true even when the peer had closed
 down and all available input had been read.
 </p>
 </li>
 <li><p> The revised HTML search system now generates better
 hyperlinks to help topics found: previously it gave problems with
 help pages with names containing e.g. spaces and slashes.
 </p>
 </li>
 <li><p> A late change in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.2 broke <span class="samp">\Sexpr</span> expressions
 in Rd files.
 </p>
 </li>
 <li><p> The creation of ticks on log axes (including by
 <code>axTicks()</code>) sometimes incorrectly omitted a tick at one end
 of the range by rounding error in a platform-dependent way.  This
 could be seen in the examples for <code>axTicks()</code>, where with
 axis limits <code>c(0.2, 88)</code> the tick for <code>0.2</code> was
 sometimes omitted.
 </p>
 </li>
 <li> <p><code>qgamma()</code> for small <code>shape</code> underflows to
 <code>0</code> rather than sometimes giving <code>NaN</code>.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=8528">PR#8528</a>, <a href="https://bugs.R-project.org/show_bug.cgi?id=14710">PR#14710</a>)
 </p>
 </li>
 <li> <p><code>mapply()</code> now gives an explicit error message (rather
 than an obscure one) if inputs of zero and positive length are
 mixed.
 </p>
 </li>
 <li><p> Setting a Hershey font family followed by string height
 query would crash R.
 </p>
 </li>
 <li> <p><code>R CMD javareconf -e</code> would fail for some shells due to
 a shift error. Also the resulting paths will no longer contain
 <code>$(JAVA_HOME)</code> as that can result in an unintended
 substitution based on &lsquo;<span class="file">Makeconf</span>&rsquo; instead of the shell
 setting.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.13.2</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>mem.limits()</code> now reports values larger than the
 maximum integer (previously documented to be reported as
 <code>NA</code>), and allows larger values to be set, including
 <code>Inf</code> to remove the limit.
 </p>
 </li>
 <li><p> The <code>print()</code> methods for classes <code>"Date"</code>,
 <code>"POSIXct"</code> and <code>"POSIXlt"</code> respect the option
 <code>"max.print"</code> and so are much faster for very long datetime
 vectors.  (Suggestion of Yohan Chalabi.)
 </p>
 </li>
 <li> <p><code>untar2()</code> now works around errors generated with
 <code>tar</code> files that use more than the standard 6 digits for the
 checksum. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14654">PR#14654</a>)
 </p>
 </li>
 <li> <p><code>install.packages()</code> with <code>Ncpus &gt; 1</code> guards
 against simultaneous installation of indirect dependencies as well
 as direct ones.
 </p>
 </li>
 <li><p> Sweave now knows about a few more Windows' encodings
 (including <code>cp1250</code> and <code>cp1257</code>) and some
 <code>inputenx</code> encodings such as <code>koi8-r</code>.
 </p>
 </li>
 <li> <p><code>postscript(colormodel = "rgb-nogray")</code> no longer sets
 the sRGB colorspace for each colour and so some viewers may render
 its files much faster than the default <code>colormodel ="rgb"</code>.
 </p>
 </li>
 <li><p> The default for <code>pdf(maxRasters=)</code> has been increased
 from 64 to 1000.
 </p>
 </li>
 <li> <p><code>readBin()</code> now warns if <code>signed = FALSE</code> is used
 inappropriately (rather than being silently ignored).
 </p>
 <p>It enforces the documented limit of <i>2^31-1</i> bytes
 in a single call.
 </p>
 </li>
 <li><p> PCRE has been updated to version 8.13, a bug-fix release
 with updated Unicode tables (version 6.0.0).  An additional patch
 (r611 from PCRE 8.20-to-be) has been added to fix a collation
 symbol recognition issue.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> It is possible to build in &lsquo;<span class="file">src/extra/xdr</span>&rsquo; on more
 platforms.  (Needed since glibc 2.14 hides its RPC implementation.)
 </p>
 </li>
 <li> <p><code>configure</code> will find the Sun TI-RPC implementation
 of xdr (in &lsquo;<span class="file">libtirpc</span>&rsquo;) provided its header files are in the
 search path: see the &lsquo;R Installation and Administration
 Manual&rsquo;.
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li><p> Using a broad <code>exportPattern</code> directive in a
 &lsquo;<span class="file">NAMESPACE</span>&rsquo; file is no longer allowed to export internal
 objects such as <code>.onLoad</code> and <code>.__S3MethodsTable__.</code> .
 </p>
 <p>These are also excluded from imports, along with <code>.First.lib</code>.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>fisher.test()</code> had a buglet: If arguments were factors
 with unused levels, levels were dropped and you would get an error
 saying that there should be at least two levels, inconsistently
 with pre-tabulated data.  (Reported by Michael Fay).
 </p>
 </li>
 <li> <p><code>package.skeleton()</code> will no longer dump S4 objects
 supplied directly rather than in a code file.  These cannot be
 restored correctly from the dumped version.
 </p>
 </li>
 <li><p> Build-time expressions in help files did not have access
 to functions in the package being built (with <code>R CMD build</code>).
 </p>
 </li>
 <li><p> Because <code>quote()</code> did not mark its result as being in
 use, modification of the result could in some circumstances
 modify the original call.
 </p>
 </li>
 <li><p> Plotting <code>pch = '.'</code> now guarantees at least a one-pixel
 dot if <code>cex &gt; 0</code>.
 </p>
 </li>
 <li><p> The very-rarely-used command-line option
 <span class="option">--max-vsize</span> was incorrectly interpreted as a number of
 Vcells and not in bytes as documented.  (Spotted by Christophe
 Rhodes.)
 </p>
 </li>
 <li><p> The HTML generated by <code>Rd2HTML()</code> comes closer to
 being standards compliant.
 </p>
 </li>
 <li> <p><code>filter(x, recursive = TRUE)</code> gave incorrect results on a
 series containing <code>NA</code>s.  (Spotted by Bill Dunlap.)
 </p>
 </li>
 <li><p> Profiling <code>stats::mle()</code> fits with a <code>fixed</code>
 parameter was not supported.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14646">PR#14646</a>)
 </p>
 </li>
 <li> <p><code>retracemem()</code> was still using positional matching.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14650">PR#14650</a>)
 </p>
 </li>
 <li><p> The <code>quantile</code> method for <code>"ecdf"</code> objects now
 works and is documented.
 </p>
 </li>
 <li> <p><code>xtabs(~ .., ..., sparse=TRUE)</code> now also works together
 with an <code>exclude = ..</code> specification.
 </p>
 </li>
 <li> <p><code>decompose()</code> computed an incorrect seasonal component for
 time series with odd frequencies.
 </p>
 </li>
 <li><p> The <code>pdf()</code> device only includes the definition of the
 sRGB colorspace in the output file for the <code>"rgb"</code> colormodel
 (and not for <code>"gray"</code> nor <code>"cmyk"</code>): this saves ca
 9KB in the output file.
 </p>
 </li>
 <li> <p><code>.hasSlot()</code> wrongly gave <code>FALSE</code> in some cases.
 </p>
 </li>
 <li> <p><code>Sweave()</code> with <code>keep.source=TRUE</code> could generate
 spurious <code>NA</code> lines when a chunk reference appeared last in
 a code chunk.
 </p>
 </li>
 <li> <p><span class="samp">\Sexpr[results=rd]</span> in an &lsquo;<span class="file">.Rd</span>&rsquo; file now first tries
 <code>parse_Rd(fragment=FALSE)</code> to allow Rd section-level macros
 to be inserted.
 </p>
 </li>
 <li><p> The <code>print()</code> method for class <code>"summary.aov"</code> did
 not pass on arguments such as <code>signif.stars</code> when
 <code>summary()</code> was called on a single object.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14684">PR#14684</a>)
 </p>
 </li>
 <li><p> In rare cases <code>ks.test()</code> could return a p-value
 very slightly less than 0 by rounding error.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14671">PR#14671</a>)
 </p>
 </li>
 <li><p> If <code>trunc()</code> was called on a <code>"POSIXlt"</code> vector
 and the result was subsetted, all but the first element was
 converted to <code>NA</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14679">PR#14679</a>)
 </p>
 </li>
 <li> <p><code>cbind()</code> and <code>rbind()</code> could cause memory
 corruption when used on a combination of raw and logical/integer
 vectors.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.13.1</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>iconv()</code> no longer translates <code>NA</code> strings as
 <code>"NA"</code>.
 </p>
 </li>
 <li> <p><code>persp(box = TRUE)</code> now warns if the surface extends
 outside the box (since occlusion for the box and axes is computed
 assuming the box is a bounding box). (<a href="https://bugs.R-project.org/show_bug.cgi?id=202">PR#202</a>)
 </p>
 </li>
 <li> <p><code>RShowDoc()</code> can now display the licences shipped with
 <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>, e.g. <code>RShowDoc("GPL-3")</code>.
 </p>
 </li>
 <li><p> New wrapper function <code>showNonASCIIfile()</code> in package
 <span class="pkg">tools</span>.
 </p>
 </li>
 <li> <p><code>nobs()</code> now has a <code>"mle"</code> method in package
 <span class="pkg">stats4</span>.
 </p>
 </li>
 <li> <p><code>trace()</code> now deals correctly with S4 reference classes
 and corresponding reference methods (e.g., <code>$trace()</code>)
 have been added.
 </p>
 </li>
 <li> <p><code>xz</code> has been updated to 5.0.3 (very minor bugfix
 release).
 </p>
 </li>
 <li> <p><code>tools::compactPDF()</code> gets more compression (usually a
 little, sometimes a lot) by using the compressed object streams of
 PDF 1.5.
 </p>
 </li>
 <li> <p><code>cairo_ps(onefile = TRUE)</code> generates encapsulated EPS
 on platforms with cairo &gt;= 1.6.
 </p>
 </li>
 <li><p> Binary reads (e.g., by <code>readChar()</code> and <code>readBin()</code>)
 are now supported on clipboard connections.  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14593">PR#14593</a>.)
 </p>
 </li>
 <li> <p><code>as.POSIXlt.factor()</code> now passes <code>...</code> to the
 character method (suggestion of Joshua Ulrich).  [Intended for <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 2.13.0 but accidentally removed before release.]
 </p>
 </li>
 <li> <p><code>vector()</code> and its wrappers such as <code>integer()</code>
 and <code>double()</code> now warn if called with a <code>length</code>
 argument of more than one element.  This helps track down user
 errors such as calling <code>double(x)</code> instead of
 <code>as.double(x)</code>.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> Building the vignette PDFs in packages <span class="pkg">grid</span> and
 <span class="pkg">utils</span> is now part of running <code>make</code> from an SVN
 checkout on a Unix-alike: a separate <code>make vignettes</code> step
 is no longer required.
 </p>
 <p>These vignettes are now made with <code>keep.source = TRUE</code> and
 hence will be laid out differently.
 </p>
 </li>
 <li> <p><code>make install-strip</code> failed under some configuration
 options.
 </p>
 </li>
 <li><p> Packages can customize non-standard installation of compiled
 code via a <code>src/install.libs.R</code> script. This allows packages that
 have architecture-specific binaries (beyond the package's shared
 objects/DLLs) to be installed in a multi-architecture setting.
 </p>
 </li></ul>


 <h4>SWEAVE &amp; VIGNETTES</h4>


 <ul>
 <li> <p><code>Sweave()</code> and <code>Stangle()</code> gain an <code>encoding</code>
 argument to specify the encoding of the vignette sources if the
 latter do not contain a <span class="samp">\usepackage[]{inputenc}</span> statement
 specifying a single input encoding.
 </p>
 </li>
 <li><p> There is a new Sweave option <code>figs.only = TRUE</code> to run
 each figure chunk only for each selected graphics device, and not
 first using the default graphics device.   This will become the
 default in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.14.0.
 </p>
 </li>
 <li><p> Sweave custom graphics devices can have a custom function
 <code>foo.off()</code> to shut them down.
 </p>
 </li>
 <li><p> Warnings are issued when non-portable filenames are found
 for graphics files (and chunks if <code>split = TRUE</code>).  Portable
 names are regarded as alphanumeric plus hyphen, underscore, plus
 and hash (periods cause problems with recognizing file
 extensions).
 </p>
 </li>
 <li><p> The <code>Rtangle()</code> driver has a new option
 <code>show.line.nos</code> which is by default false; if true it
 annotates code chunks with a comment giving the line number of the
 first line in the sources (the behaviour of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> &gt;= 2.12.0).
 </p>
 </li>
 <li><p> Package installation tangles the vignette sources: this step
 now converts the vignette sources from the vignette/package
 encoding to the current encoding, and records the encoding (if not
 ASCII) in a comment line at the top of the installed &lsquo;<span class="file">.R</span>&rsquo;
 file.
 </p>
 </li></ul>


 <h4>LICENCE</h4>


 <ul>
 <li><p> No parts of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> are now licensed solely under GPL-2.  The
 licences for packages <a href="https://CRAN.R-project.org/package=rpart"><span class="pkg">rpart</span></a> and <a href="https://CRAN.R-project.org/package=survival"><span class="pkg">survival</span></a> have been
 changed, which means that the licence terms for <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> as distributed
 are GPL-2 | GPL-3.
 </p>
 </li></ul>


 <h4>DEPRECATED AND DEFUNCT</h4>


 <ul>
 <li><p> The internal functions <code>.readRDS()</code> and
 <code>.saveRDS()</code> are now deprecated in favour of the public
 functions <code>readRDS()</code> and <code>saveRDS()</code> introduced in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 2.13.0.
 </p>
 </li>
 <li><p> Switching off lazy-loading of code <em>via</em> the
 <span class="samp">LazyLoad</span> field of the &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file is now
 deprecated.  In future all packages will be lazy-loaded.
 </p>
 </li>
 <li><p> The off-line <code>help()</code> types <span class="samp">"postscript"</span> and
 <span class="samp">"ps"</span> are deprecated.
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li> <p><code>R CMD check</code> on a multi-architecture installation now
 skips the user's &lsquo;<span class="file">.Renviron</span>&rsquo; file for the
 architecture-specific tests (which do read the
 architecture-specific &lsquo;<span class="file">Renviron.site</span>&rsquo; files).  This is
 consistent with single-architecture checks, which use
 <span class="option">--no-environ</span>.
 </p>
 </li>
 <li> <p><code>R CMD build</code> now looks for &lsquo;<span class="file">DESCRIPTION</span>&rsquo;
 fields <span class="samp">BuildResaveData</span> and <span class="samp">BuildKeepEmpty</span> for
 per-package overrides.  See &lsquo;Writing R Extensions&rsquo;.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>plot.lm(which = 5)</code> was intended to order factor
 levels in increasing order of mean standardized residual.   It
 ordered the factor labels correctly, but could plot the wrong
 group of residuals against the label.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14545">PR#14545</a>)
 </p>
 </li>
 <li> <p><code>mosaicplot()</code> could clip the factor labels, and could
 overlap them with the cells if a non-default value of
 <code>cex.axis</code> was used.  (Related to <a href="https://bugs.R-project.org/show_bug.cgi?id=14550">PR#14550</a>.)
 </p>
 </li>
 <li> <p><code>dataframe[[row,col]]</code> now dispatches on <code>[[</code>
 methods for the selected column.  (Spotted by Bill Dunlap).
 </p>
 </li>
 <li> <p><code>sort.int()</code> would strip the class of an object, but
 leave its object bit set.  (Reported by Bill Dunlap.)
 </p>
 </li>
 <li> <p><code>pbirthday()</code> and <code>qbirthday()</code> did not implement
 the algorithm exactly as given in their reference and so were
 unnecessarily inaccurate.
 </p>
 <p><code>pbirthday()</code> now solves the approximate formula analytically
 rather than using <code>uniroot()</code> on a discontinuous function.
 </p>
 <p>The description of the problem was inaccurate: the probability is
 a tail probability (&lsquo;2 <em>or more</em> people share a birthday&rsquo;)
 </p>
 </li>
 <li><p> Complex arithmetic sometimes warned incorrectly about
 producing NAs when there were NaNs in the input.
 </p>
 </li>
 <li> <p><code>seek(origin = "current")</code> incorrectly reported it was
 not implemented for a <code>gzfile()</code> connection.
 </p>
 </li>
 <li> <p><code>c()</code>, <code>unlist()</code>, <code>cbind()</code> and
 <code>rbind()</code> could silently overflow the maximum vector length
 and cause a segfault.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14571">PR#14571</a>)
 </p>
 </li>
 <li><p> The <code>fonts</code> argument to <code>X11(type = "Xlib")</code> was
 being ignored.
 </p>
 </li>
 <li><p> Reading (e.g., with <code>readBin()</code>) from a raw connection
 was not advancing the pointer, so successive reads would read the
 same value.  (Spotted by Bill Dunlap.)
 </p>
 </li>
 <li><p> Parsed text containing embedded newlines was printed
 incorrectly by <code>as.character.srcref()</code>.  (Reported by Hadley
 Wickham.)
 </p>
 </li>
 <li> <p><code>decompose()</code> used with a series of a non-integer
 number of periods returned a seasonal component shorter than the
 original series.  (Reported by Rob Hyndman.)
 </p>
 </li>
 <li> <p><code>fields = list()</code> failed for <code>setRefClass()</code>.
 (Reported by Michael Lawrence.)
 </p>
 </li>
 <li><p> Reference classes could not redefine an inherited field
 which had class <code>"ANY"</code>. (Reported by Janko Thyson.)
 </p>
 </li>
 <li><p> Methods that override previously loaded versions will now
 be installed and called.  (Reported by Iago Mosqueira.)
 </p>
 </li>
 <li> <p><code>addmargins()</code> called <code>numeric(apos)</code> rather than
 <code>numeric(length(apos))</code>.
 </p>
 </li>
 <li><p> The HTML help search sometimes produced bad links.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14608">PR#14608</a>)
 </p>
 </li>
 <li><p> Command completion will no longer be broken if
 <code>tail.default()</code> is redefined by the user. (Problem reported
 by Henrik Bengtsson.)
 </p>
 </li>
 <li><p> LaTeX rendering of markup in titles of help pages has been
 improved; in particular, <code>\eqn{}</code> may be used there.
 </p>
 </li>
 <li> <p><code>isClass()</code> used its own namespace as the default of
 the <code>where</code> argument inadvertently.
 </p>
 </li>
 <li><p> Rd conversion to latex mishandled multi-line titles
 (including cases where there was a blank line in the <span class="samp">\title</span>
 section).  (It seems this happened only in 2.13.0 patched.)
 </p>
 </li>
 <li> <p><code>postscript()</code> with an sRGB colormodel now uses sRGB
 for raster images (in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.[01] it used uncalibrated RGB).
 </p>
 <p>There is no longer an undocumented 21845-pixel limit on raster images.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.13.0</h3>


 <h4>SIGNIFICANT USER-VISIBLE CHANGES</h4>


 <ul>
 <li> <p><code>replicate()</code> (by default) and <code>vapply()</code> (always)
 now return a higher-dimensional array instead of a matrix in the case
 where the inner function value is an array of dimension &gt;= 2.
 </p>
 </li>
 <li><p> Printing and formatting of floating point numbers is now
 using the correct number of digits, where it previously rarely
 differed by a few digits. (See &ldquo;scientific&rdquo; entry below.)
 This affects <em>many</em> &lsquo;<span class="file">*.Rout.save</span>&rsquo; checks in packages.
 </p>
 </li></ul>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>normalizePath()</code> has been moved to the <span class="pkg">base</span>
 package (from <span class="pkg">utils</span>): this is so it can be used by
 <code>library()</code> and friends.
 </p>
 <p>It now does tilde expansion.
 </p>
 <p>It gains new arguments <code>winslash</code> (to select the separator on
 Windows) and <code>mustWork</code> to control the action if a canonical
 path cannot be found.
 </p>
 </li>
 <li><p> The previously barely documented limit of 256 bytes on a
 symbol name has been raised to 10,000 bytes (a sanity check).
 Long symbol names can sometimes occur when deparsing expressions
 (for example, in <code>model.frame</code>).
 </p>
 </li>
 <li> <p><code>reformulate()</code> gains a <code>intercept</code> argument.
 </p>
 </li>
 <li> <p><code>cmdscale(add = FALSE)</code> now uses the more common
 definition that there is a representation in <code>n-1</code> or less
 dimensions, and only dimensions corresponding to positive
 eigenvalues are used.  (Avoids confusion such as <a href="https://bugs.R-project.org/show_bug.cgi?id=14397">PR#14397</a>.)
 </p>
 </li>
 <li><p> Names used by <code>c()</code>, <code>unlist()</code>, <code>cbind()</code>
 and <code>rbind()</code> are marked with an encoding when this can be
 ascertained.
 </p>
 </li>
 <li><p> R colours are now defined to refer to the sRGB color space.
 </p>
 <p>The PDF, PostScript, and Quartz graphics devices record this fact.
 X11 (and cairo) and Windows just assume that your screen conforms.
 </p>
 </li>
 <li> <p><code>system.file()</code> gains a <code>mustWork</code> argument
 (suggestion of Bill Dunlap).
 </p>
 </li>
 <li> <p><code>new.env(hash = TRUE)</code> is now the default.
 </p>
 </li>
 <li> <p><code>list2env(envir = NULL)</code> defaults to hashing (with a
 suitably sized environment) for lists of more than 100 elements.
 </p>
 </li>
 <li> <p><code>text()</code> gains a formula method.
 </p>
 </li>
 <li> <p><code>IQR()</code> now has a <code>type</code> argument which is passed
 to <code>quantile()</code>.
 </p>
 </li>
 <li> <p><code>as.vector()</code>, <code>as.double()</code> etc duplicate less
 when they leave the mode unchanged but remove attributes.
 </p>
 <p><code>as.vector(mode = "any")</code> no longer duplicates when it does
 not remove attributes.  This helps memory usage in <code>matrix()</code>
 and <code>array()</code>.
 </p>
 <p><code>matrix()</code> duplicates less if <code>data</code> is an atomic vector
 with attributes such as names (but no class).
 </p>
 <p><code>dim(x) &lt;- NULL</code> duplicates less if <code>x</code> has neither
 dimensions nor names (since this operation removes names and
 dimnames).
 </p>
 </li>
 <li> <p><code>setRepositories()</code> gains an <code>addURLs</code> argument.
 </p>
 </li>
 <li> <p><code>chisq.test()</code> now also returns a <code>stdres</code>
 component, for standardized residuals (which have unit variance,
 unlike the Pearson residuals).
 </p>
 </li>
 <li> <p><code>write.table()</code> and friends gain a <code>fileEncoding</code>
 argument, to simplify writing files for use on other OSes (e.g., a
 spreadsheet intended for Windows or Mac OS X Excel).
 </p>
 </li>
 <li><p> Assignment expressions of the form <code>foo::bar(x) &lt;- y</code>
 and <code>foo:::bar(x) &lt;- y</code> now work; the replacement functions
 used are <code>foo::`bar&lt;-`</code> and <code>foo:::`bar&lt;-`</code>.
 </p>
 </li>
 <li> <p><code>Sys.getenv()</code> gains a <code>names</code> argument so
 <code>Sys.getenv(x, names = FALSE)</code> can replace the common idiom
 of <code>as.vector(Sys.getenv())</code>.  The default has been changed
 to not name a length-one result.
 </p>
 </li>
 <li><p> Lazy loading of environments now preserves attributes and
 locked status. (The locked status of bindings and active bindings
 are still not preserved; this may be addressed in the future).
 </p>
 </li>
 <li> <p><code>options("install.lock")</code> may be set to <code>FALSE</code>
 so that <code>install.packages()</code> defaults to <span class="option">--no-lock</span>
 installs, or (on Windows) to <code>TRUE</code> so that binary installs
 implement locking.
 </p>
 </li>
 <li> <p><code>sort(partial = p)</code> for large <code>p</code> now tries
 Shellsort if quicksort is not appropriate and so works for
 non-numeric atomic vectors.
 </p>
 </li>
 <li> <p><code>sapply()</code> gets a new option <code>simplify = "array"</code>
 which returns a &ldquo;higher rank&rdquo; array instead of just a matrix
 when <code>FUN()</code> returns a <code>dim()</code> length of two or more.
 </p>
 <p><code>replicate()</code> has this option set by default, and
 <code>vapply()</code> now behaves that way internally.
 </p>
 </li>
 <li> <p><code>aperm()</code> becomes S3 generic and gets a <code>table</code>
 method which preserves the class.
 </p>
 </li>
 <li> <p><code>merge()</code> and <code>as.hclust()</code> methods for objects of
 class <code>"dendrogram"</code> are now provided.
 </p>
 </li>
 <li><p> The character method of <code>as.POSIXlt()</code> now tries to
 find a format that works for all non-<code>NA</code> inputs, not just
 the first one.
 </p>
 </li>
 <li> <p><code>str()</code> now has a method for class <code>"Date"</code>
 analogous to that for class <code>"POSIXt"</code>.
 </p>
 </li>
 <li><p> New function <code>file.link()</code> to create hard links on
 those file systems (POSIX, NTFS but not FAT) that support them.
 </p>
 </li>
 <li><p> New <code>Summary()</code> group method for class <code>"ordered"</code>
 implements <code>min()</code>, <code>max()</code> and <code>range()</code> for
 ordered factors.
 </p>
 </li>
 <li> <p><code>mostattributes&lt;-()</code> now consults the <code>"dim"</code>
 attribute and not the <code>dim()</code> function, making it more useful
 for objects (such as data frames) from classes with methods for
 <code>dim()</code>.  It also uses <code>attr&lt;-()</code> in preference to the
 generics <code>name&lt;-()</code>, <code>dim&lt;-()</code> and <code>dimnames&lt;-()</code>.
 (Related to <a href="https://bugs.R-project.org/show_bug.cgi?id=14469">PR#14469</a>.)
 </p>
 </li>
 <li><p> There is a new option <code>"browserNLdisabled"</code> to disable
 the use of an empty (e.g., via the &lsquo;Return&rsquo; key) as a synonym
 for <code>c</code> in <code>browser()</code> or <code>n</code> under <code>debug()</code>.
 (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14472">PR#14472</a>.)
 </p>
 </li>
 <li> <p><code>example()</code> gains optional new arguments
 <code>character.only</code> and <code>give.lines</code> enabling programmatic
 exploration.
 </p>
 </li>
 <li> <p><code>serialize()</code> and <code>unserialize()</code> are no longer
 described as &lsquo;experimental&rsquo;.  The interface is now regarded
 as stable, although the serialization format may well change in
 future releases.  (<code>serialize()</code> has a new argument
 <code>version</code> which would allow the current format to be written
 if that happens.)
 </p>
 <p>New functions <code>saveRDS()</code> and <code>readRDS()</code> are public
 versions of the &lsquo;internal&rsquo; functions <code>.saveRDS()</code> and
 <code>.readRDS()</code> made available for general use.  The dot-name
 versions remain available as several package authors have made use
 of them, despite the documentation.  <code>readRDS()</code> no longer
 wraps non-file connections in a call to <code>gzcon()</code>, for
 efficiency (see its documentation).
 </p>
 <p><code>saveRDS()</code> supports <code>compress = "xz"</code>.
 </p>
 </li>
 <li><p> Many functions when called with a not-open connection will
 now ensure that the connection is left not-open in the event of
 error.  These include <code>read.dcf()</code>, <code>dput()</code>,
 <code>dump()</code>, <code>load()</code>, <code>parse()</code>, <code>readBin()</code>,
 <code>readChar()</code>, <code>readLines()</code>, <code>save()</code>,
 <code>writeBin()</code>, <code>writeChar()</code>, <code>writeLines()</code>,
 <code>.readRDS()</code>, <code>.saveRDS()</code> and <code>tools::parse_Rd()</code>,
 as well as functions calling these.
 </p>
 </li>
 <li><p> Public functions <code>find.package()</code> and <code>path.package()</code>
 replace the internal dot-name versions.
 </p>
 </li>
 <li><p> The default method for <code>terms()</code> now looks for a
 <code>"terms"</code> attribute if it does not find a <code>"terms"</code>
 component, and so works for model frames.
 </p>
 </li>
 <li> <p><code>httpd()</code> handlers receive an additional argument
 containing the full request headers as a raw vector (this can be
 used to parse cookies, multi-part forms etc.). The recommended
 full signature for handlers is therefore <code>function(url,
 	query, body, headers, ...)</code>.
 </p>
 </li>
 <li> <p><code>file.edit()</code> gains a <code>fileEncoding</code> argument to
 specify the encoding of the file(s).
 </p>
 </li>
 <li><p> The format of the HTML package listings has changed.  If
 there is more than one library tree , a table of links to
 libraries is provided at the top and bottom of the page.  Where a
 library contains more than 100 packages, an alphabetic index is
 given at the top of the section for that library.  (As a
 consequence, package names are now sorted case-insensitively
 whatever the locale.)
 </p>
 </li>
 <li> <p><code>isSeekable()</code> now returns <code>FALSE</code> on connections
 which have non-default <code>encoding</code>.  Although documented to
 record if &lsquo;in principle&rsquo; the connection supports seeking,
 it seems safer to report <code>FALSE</code> when it may not work.
 </p>
 </li>
 <li> <p><code>R CMD REMOVE</code> and <code>remove.packages()</code> now
 remove file <code>R.css</code> when removing all remaining packages in a
 library tree.  (Related to the wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14475">PR#14475</a>: note that this
 file is no longer installed.)
 </p>
 </li>
 <li> <p><code>unzip()</code> now has a <code>unzip</code> argument like
 <code>zip.file.extract()</code>.  This allows an external
 <code>unzip</code> program to be used, which can be useful to access
 features supported by Info-ZIP's <code>unzip</code> version 6 which
 is now becoming more widely available.
 </p>
 </li>
 <li><p> There is a simple <code>zip()</code> function, as wrapper for an
 external <code>zip</code> command.
 </p>
 </li>
 <li> <p><code>bzfile()</code> connections can now read from concatenated
 <code>bzip2</code> files (including files written with
 <code>bzfile(open = "a")</code>) and files created by some other
 compressors (such as the example of <a href="https://bugs.R-project.org/show_bug.cgi?id=14479">PR#14479</a>).
 </p>
 </li>
 <li><p> The primitive function <code>c()</code> is now of type BUILTIN.
 </p>
 </li>
 <li> <p><code>plot(&lt;dendrogram&gt;, .., nodePar=*)</code> now obeys an optional
 <code>xpd</code> specification (allowing clipping to be turned off
 completely).
 </p>
 </li>
 <li> <p><code>nls(algorithm="port")</code> now shares more code with
 <code>nlminb()</code>, and is more consistent with the other
 <code>nls()</code> algorithms in its return value.
 </p>
 </li>
 <li> <p><code>xz</code> has been updated to 5.0.1 (very minor bugfix
 release).
 </p>
 </li>
 <li> <p><code>image()</code> has gained a logical <code>useRaster</code>
 argument allowing it to use a bitmap raster for plotting a
 regular grid instead of polygons. This can be more efficient, but
 may not be supported by all devices. The default is <code>FALSE</code>.
 </p>
 </li>
 <li> <p><code>list.files()</code>/<code>dir()</code> gains a new argument
 <code>include.dirs()</code> to include directories in the listing when
 <code>recursive = TRUE</code>.
 </p>
 </li>
 <li><p> New function <code>list.dirs()</code> lists all directories,
 (even empty ones).
 </p>
 </li>
 <li> <p><code>file.copy()</code> now (by default) copies
 read/write/execute permissions on files, moderated by the current
 setting of <code>Sys.umask()</code>.
 </p>
 </li>
 <li> <p><code>Sys.umask()</code> now accepts <code>mode = NA</code> and returns
 the current <code>umask</code> value (visibly) without changing it.
 </p>
 </li>
 <li><p> There is a <code>!</code> method for classes <code>"octmode"</code> and
 <code>"hexmode"</code>: this allows <code>xor(a, b)</code> to work if both
 <code>a</code> and <code>b</code> are from one of those classes.
 </p>
 </li>
 <li> <p><code>as.raster()</code> no longer fails for vectors or matrices
 containing <code>NA</code>s.
 </p>
 </li>
 <li><p> New hook <code>"before.new.plot"</code> allows functions to be run
 just before advancing the frame in <code>plot.new</code>, which is
 potentially useful for custom figure layout implementations.
 </p>
 </li>
 <li><p> Package <span class="pkg">tools</span> has a new function <code>compactPDF()</code>
 to try to reduce the size of PDF files <em>via</em> <code>qpdf</code>
 or <code>gs</code>.
 </p>
 </li>
 <li> <p><code>tar()</code> has a new argument <code>extra_flags</code>.
 </p>
 </li>
 <li> <p><code>dotchart()</code> accepts more general objects <code>x</code> such
 as 1D tables which can be coerced by <code>as.numeric()</code> to a
 numeric vector, with a warning since that might not be
 appropriate.
 </p>
 </li>
 <li><p> The previously internal function <code>create.post()</code> is now
 exported from <span class="pkg">utils</span>, and the documentation for
 <code>bug.report()</code> and <code>help.request()</code> now refer to that
 for <code>create.post()</code>.
 </p>
 <p>It has a new <code>method = "mailto"</code> on Unix-alikes similar to
 that on Windows: it invokes a default mailer via <code>open</code>
 (Mac OS X) or <code>xdg-open</code> or the default browser (elsewhere).
 </p>
 <p>The default for <code>ccaddress</code> is now
 <code>getOption("ccaddress")</code> which is by default unset: using the
 username as a mailing address nowadays rarely works as expected.
 </p>
 </li>
 <li><p> The default for <code>options("mailer")</code> is now
 <code>"mailto"</code> on all platforms.
 </p>
 </li>
 <li> <p><code>unlink()</code> now does tilde-expansion (like most other
 file functions).
 </p>
 </li>
 <li> <p><code>file.rename()</code> now allows vector arguments (of the
 same length).
 </p>
 </li>
 <li><p> The <code>"glm"</code> method for <code>logLik()</code> now returns an
 <code>"nobs"</code> attribute (which <code>stats4::BIC()</code> assumed it
 did).
 </p>
 <p>The <code>"nls"</code> method for <code>logLik()</code> gave incorrect results
 for zero weights.
 </p>
 </li>
 <li><p> There is a new generic function <code>nobs()</code> in package
 <span class="pkg">stats</span>, to extract from model objects a suitable value for
 use in BIC calculations.  An S4 generic derived from it is defined
 in package <span class="pkg">stats4</span>.
 </p>
 </li>
 <li><p> Code for S4 reference-class methods is now examined for
 possible errors in non-local assignments.
 </p>
 </li>
 <li> <p><code>findClasses</code>, <code>getGeneric</code>, <code>findMethods</code>
 and <code>hasMethods</code> are revised to deal consistently with
 the <code>package=</code> argument and be consistent with soft namespace
 policy for finding objects.
 </p>
 </li>
 <li> <p><code>tools::Rdiff()</code> now has the option to return not only
 the status but a character vector of observed differences (which
 are still by default sent to &lsquo;<span class="file">stdout</span>&rsquo;).
 </p>
 </li>
 <li><p> The startup environment variables <span class="env">R_ENVIRON_USER</span>,
 <span class="env">R_ENVIRON</span>, <span class="env">R_PROFILE_USER</span> and <span class="env">R_PROFILE</span> are now
 treated more consistently.  In all cases an empty value is
 considered to be set and will stop the default being used, and for
 the last two tilde expansion is performed on the file name.  (Note
 that setting an empty value is probably impossible on Windows.)
 </p>
 </li>
 <li><p> Using <code>R --no-environ CMD</code>, <code>R
 	--no-site-file CMD</code> or <code>R --no-init-file CMD</code> sets
 environment variables so these settings are passed on to child <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 processes, notably those run by <code>INSTALL</code>, <code>check</code> and
 <code>build</code>. <code>R --vanilla CMD</code> sets these three options
 (but not <span class="option">--no-restore</span>).
 </p>
 </li>
 <li> <p><code>smooth.spline()</code> is somewhat faster.  With
 <code>cv=NA</code> it allows some leverage computations to be skipped,
 </p>
 </li>
 <li><p> The internal (C) function <span class="samp">scientific()</span>, at the heart
 of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>'s <code>format.info(x)</code>, <code>format(x)</code>, <code>print(x)</code>,
 etc, for numeric <code>x</code>, has been re-written in order to provide
 slightly more correct results, fixing <a href="https://bugs.R-project.org/show_bug.cgi?id=14491">PR#14491</a>, notably in
 border cases including when <code>digits &gt;= 16</code>, thanks to
 substantial contributions (code and experiments) from Petr
 Savicky.  This affects a noticeable amount of numeric output from
 <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 </p>
 </li>
 <li><p> A new function <code>grepRaw()</code> has been introduced for
 finding subsets of raw vectors. It supports both literal searches
 and regular expressions.
 </p>
 </li>
 <li><p> Package <span class="pkg">compiler</span> is now provided as a standard
 package.  See <code>?compiler::compile</code> for information on how to
 use the compiler.  This package implements a byte code compiler
 for <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>: by default the compiler is not used in this release.  See
 the &lsquo;R Installation and Administration Manual&rsquo; for how to
 compile the base and recommended packages.
 </p>
 </li>
 <li><p> Providing an <code>exportPattern</code> directive in a NAMESPACE
 file now causes classes to be exported according to the same
 pattern, for example the default from <code>package.skeleton()</code>
 to specify all names starting with a letter.  An explicit
 directive to <code>exportClassPattern</code> will still over-ride.
 </p>
 </li>
 <li><p> There is an additional marked encoding <code>"bytes"</code> for
 character strings.  This is intended to be used for non-ASCII
 strings which should be treated as a set of bytes, and never
 re-encoded as if they were in the encoding of the current locale:
 <code>useBytes = TRUE</code> is automatically selected in functions such
 as <code>writeBin()</code>, <code>writeLines()</code>, <code>grep()</code> and
 <code>strsplit()</code>.
 </p>
 <p>Only a few character operations are supported (such as <code>substr()</code>).
 </p>
 <p>Printing, <code>format()</code> and <code>cat()</code> will represent non-ASCII
 bytes in such strings by a <span class="samp">\xab</span> escape.
 </p>
 </li>
 <li><p> The new function <code>removeSource()</code> removes the internally
 stored source from a function.
 </p>
 </li>
 <li> <p><code>"srcref"</code> attributes now include two additional
 line number values, recording the line numbers in the order they
 were parsed.
 </p>
 </li>
 <li><p> New functions have been added for source reference access:
 <code>getSrcFilename()</code>, <code>getSrcDirectory()</code>,
 <code>getSrcLocation()</code> and <code>getSrcref()</code>.
 </p>
 </li>
 <li> <p><code>Sys.chmod()</code> has an extra argument <code>use_umask</code>
 which defaults to true and restricts the file mode by the current
 setting of <code>umask</code>.  This means that all the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> functions
 which manipulate file/directory permissions by default respect
 <code>umask</code>, notably <code>R CMD INSTALL</code>.
 </p>
 </li>
 <li> <p><code>tempfile()</code> has an extra argument <code>fileext</code> to
 create a temporary filename with a specified extension.
 (Suggestion and initial implementation by Dirk Eddelbuettel.)
 </p>
 <p>There are improvements in the way <code>Sweave()</code> and
 <code>Stangle()</code> handle non-ASCII vignette sources, especially in
 a UTF-8 locale: see &lsquo;Writing R Extensions&rsquo; which now has a
 subsection on this topic.
 </p>
 </li>
 <li> <p><code>factanal()</code> now returns the rotation matrix if a
 rotation such as <code>"promax"</code> is used, and hence factor
 correlations are displayed.  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=12754">PR#12754</a>.)
 </p>
 </li>
 <li><p> The <code>gctorture2()</code> function provides a more refined
 interface to the GC torture process.  Environment variables
 <span class="env">R_GCTORTURE</span>, <span class="env">R_GCTORTURE_WAIT</span>, and
 <span class="env">R_GCTORTURE_INHIBIT_RELEASE</span> can also be used to control the
 GC torture process.
 </p>
 </li>
 <li> <p><code>file.copy(from, to)</code> no longer regards it as an error
 to supply a zero-length <code>from</code>: it now simply does nothing.
 </p>
 </li>
 <li> <p><code>rstandard.glm()</code> gains a <code>type</code> argument which can
 be used to request standardized Pearson residuals.
 </p>
 </li>
 <li><p> A start on a Turkish translation, thanks to Murat Alkan.
 </p>
 </li>
 <li> <p><code>.libPaths()</code> calls <code>normalizePath(winslash = "/")</code>
 on the paths: this helps (usually) to present them in a
 user-friendly form and should detect duplicate paths accessed via
 different symbolic links.
 </p>
 </li>
 <li> <p><code>download.file()</code> can be now used with external methods
 even if there are spaces in the URL or the target filename.
 </p>
 </li>
 <li><p> The first formal argument of <code>.C()</code>, <code>.Fortran()</code>,
 <code>.Call()</code> and <code>.External()</code> is now referred to a
 <code>.NAME</code> rather than <code>name</code>: however actual calls should
 have an unnamed first argument.
 </p>
 </li></ul>


 <h4>SWEAVE CHANGES</h4>


 <ul>
 <li> <p><code>Sweave()</code> has options to produce PNG and JPEG figures,
 and to use a custom function to open a graphics device (see
 <code>?RweaveLatex</code>).<br />
 (Based in part on the contribution of <a href="https://bugs.R-project.org/show_bug.cgi?id=14418">PR#14418</a>.)
 </p>
 </li>
 <li><p> The default for <code>Sweave()</code> is to produce only PDF
 figures (rather than both EPS and PDF).
 </p>
 </li>
 <li><p> Environment variable <span class="env">SWEAVE_OPTIONS</span> can be used to
 supply defaults for existing or new options to be applied after
 the Sweave driver setup has been run.
 </p>
 </li>
 <li><p> The Sweave manual is now included as a vignette in the
 <span class="pkg">utils</span> package.
 </p>
 </li>
 <li> <p><code>Sweave()</code> handles <code>keep.source=TRUE</code> much better:
 it could duplicate some lines and omit comments. (Reported by John
 Maindonald and others.)
 </p>
 </li></ul>


 <h4>C-LEVEL FACILITIES</h4>


 <ul>
 <li><p> Because they use a C99 interface which a C++ compiler is not
 required to support, <code>Rvprintf</code> and <code>REvprintf</code> are only
 defined by &lsquo;<span class="file">R_ext/Print.h</span>&rsquo; in C++ code if the macro
 <code>R_USE_C99_IN_CXX</code> is defined when it is included.
 </p>
 </li>
 <li> <p><code>pythag</code> duplicated the C99 function <code>hypot</code>.  It
 is no longer provided, but is used as a substitute for <code>hypot</code>
 in the very unlikely event that the latter is not available.
 </p>
 </li>
 <li> <p><code>R_inspect(obj)</code> and <code>R_inspect3(obj, deep, pvec)</code>
 are (hidden) C-level entry points to the internal <code>inspect</code>
 function and can be used for C-level debugging (e.g., in conjunction
 with the <code>p</code> command in <code>gdb</code>).
 </p>
 </li>
 <li><p> Compiling R with <span class="option">--enable-strict-barrier</span> now also
 enables additional checking for use of unprotected objects. In
 combination with <code>gctorture()</code> or <code>gctorture2()</code> and a
 C-level debugger this can be useful for tracking down memory
 protection issues.
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li> <p><code>R CMD Rdiff</code> is now implemented in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> on
 Unix-alikes (as it has been on Windows since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.12.0).
 </p>
 </li>
 <li> <p><code>R CMD build</code> no longer does any cleaning in the
 supplied package directory: all the cleaning is done in the copy.
 </p>
 <p>It has a new option <span class="option">--install-args</span> to pass arguments to
 <code>R CMD INSTALL</code> for <span class="option">--build</span> (but not when
 installing to rebuild vignettes).
 </p>
 <p>There is new option, <span class="option">--resave-data</span>, to call
 <code>tools::resaveRdaFiles()</code> on the &lsquo;<span class="file">data</span>&rsquo; directory, to
 compress tabular files (&lsquo;<span class="file">.tab</span>&rsquo;, &lsquo;<span class="file">.csv</span>&rsquo; etc) and to
 convert &lsquo;<span class="file">.R</span>&rsquo; files to &lsquo;<span class="file">.rda</span>&rsquo; files.  The default,
 <span class="option">--resave-data=gzip</span>, is to do so in a way compatible even
 with years-old versions of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>, but better compression is given by
 <span class="option">--resave-data=best</span>, requiring <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> &gt;= 2.10.0.
 </p>
 <p>It now adds a &lsquo;<span class="file">datalist</span>&rsquo; file for &lsquo;<span class="file">data</span>&rsquo; directories of
 more than 1Mb.
 </p>
 <p>Patterns in &lsquo;<span class="file">.Rbuildignore</span>&rsquo; are now also matched against all
 directory names (including those of empty directories).
 </p>
 <p>There is a new option, <span class="option">--compact-vignettes</span>, to try
 reducing the size of PDF files in the &lsquo;<span class="file">inst/doc</span>&rsquo; directory.
 Currently this tries <code>qpdf</code>: other options may be used in
 future.
 </p>
 <p>When re-building vignettes and a &lsquo;<span class="file">inst/doc/Makefile</span>&rsquo; file is
 found, <code>make clean</code> is run if the makefile has a
 <code>clean:</code> target.
 </p>
 <p>After re-building vignettes the default clean-up operation will
 remove any directories (and not just files) created during the
 process: e.g., one package created a &lsquo;<span class="file">.R_cache</span>&rsquo; directory.
 </p>
 <p>Empty directories are now removed unless the option
 <span class="option">--keep-empty-dirs</span> is given (and a few packages do
 deliberately include empty directories).
 </p>
 <p>If there is a field <code>BuildVignettes</code> in the package
 &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file with a false value, re-building the
 vignettes is skipped.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now also checks for filenames that are
 case-insensitive matches to Windows' reserved file names with
 extensions, such as &lsquo;<span class="file">nul.Rd</span>&rsquo;, as these have caused problems
 on some Windows systems.
 </p>
 <p>It checks for inefficiently saved &lsquo;<span class="file">data/*.rda</span>&rsquo; and
 &lsquo;<span class="file">data/*.RData</span>&rsquo; files, and reports on those large than 100Kb.
 A more complete check (including of the type of compression, but
 potentially much slower) can be switched on by setting environment
 variable <span class="env">_R_CHECK_COMPACT_DATA2_</span> to <span class="samp">TRUE</span>.
 </p>
 <p>The types of files in the &lsquo;<span class="file">data</span>&rsquo; directory are now checked,
 as packages are <em>still</em> misusing it for non-<span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> data files.
 </p>
 <p>It now extracts and runs the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> code for each vignette in a
 separate directory and <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> process: this is done in the package's
 declared encoding.  Rather than call
 <code>tools::checkVignettes()</code>, it calls
 <code>tools::buildVignettes()</code> to see if the vignettes can be
 re-built as they would be by <code>R CMD build</code>.  Option
 <span class="option">--use-valgrind</span> now applies only to these runs, and not
 when running code to rebuild the vignettes.  This version does a
 much better job of suppressing output from successful vignette
 tests.
 </p>
 <p>The &lsquo;<span class="file">00check.log</span>&rsquo; file is a more complete record of what is
 output to &lsquo;<span class="file">stdout</span>&rsquo;: in particular contains more details of
 the tests.
 </p>
 <p>It now checks all syntactically valid Rd usage entries, and warns
 about assignments (unless these give the usage of replacement
 functions).
 </p>
 <p>&lsquo;<span class="file">.tar.xz</span>&rsquo; compressed tarballs are now allowed, if
 <code>tar</code> supports them (and setting environment variable
 <span class="env">TAR</span> to <span class="samp">internal</span> ensures so on all platforms).
 </p>
 </li>
 <li> <p><code>R CMD check</code> now warns if it finds
 &lsquo;<span class="file">inst/doc/makefile</span>&rsquo;, and <code>R CMD build</code> renames such a
 file to &lsquo;<span class="file">inst/doc/Makefile</span>&rsquo;.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> Installing <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> no longer tries to find <code>perl</code>, and
 <code>R CMD</code> no longer tries to substitute a full path for
 <code>awk</code> nor <code>perl</code> &ndash; this was a legacy from the
 days when they were used by <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> itself.  Because a couple of
 packages do use <code>awk</code>, it is set as the <code>make</code>
 (rather than environment) variable <span class="env">AWK</span>.
 </p>
 </li>
 <li> <p><code>make check</code> will now fail if there are differences
 from the reference output when testing package examples and if
 environment variable <span class="env">R_STRICT_PACKAGE_CHECK</span> is set to a true
 value.
 </p>
 </li>
 <li><p> The C99 double complex type is now required.
 </p>
 <p>The C99 complex trigonometric functions (such as <span class="samp">csin</span>) are
 not currently required (FreeBSD lacks most of them): substitutes
 are used if they are missing.
 </p>
 </li>
 <li><p> The C99 system call <span class="samp">va_copy</span> is now required.
 </p>
 </li>
 <li><p> If environment variable <span class="env">R_LD_LIBRARY_PATH</span> is set
 during configuration (for example in &lsquo;<span class="file">config.site</span>&rsquo;) it is
 used unchanged in file &lsquo;<span class="file">etc/ldpaths</span>&rsquo; rather than being
 appended to.
 </p>
 </li>
 <li> <p><code>configure</code> looks for support for OpenMP and if
 found compiles <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> with appropriate flags and also makes them
 available for use in packages: see &lsquo;Writing R Extensions&rsquo;.
 </p>
 <p>This is currently experimental, and is only used in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> with a
 single thread for <code>colSums()</code> and <code>colMeans()</code>.  Expect
 it to be more widely used in later versions of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 </p>
 <p>This can be disabled by the <span class="option">--disable-openmp</span> flag.
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li> <p><code>R CMD INSTALL --clean</code> now removes copies of a
 &lsquo;<span class="file">src</span>&rsquo; directory which are created when multiple
 sub-architectures are in use.  (Following a comment from
 Berwin Turlach.)
 </p>
 </li>
 <li><p> File &lsquo;<span class="file">R.css</span>&rsquo; is now installed on a per-package basis
 (in the package's &lsquo;<span class="file">html</span>&rsquo; directory) rather than in each
 library tree, and this is used for all the HTML pages in the
 package.  This helps when installing packages with static HTML
 pages for use on a webserver.  It will also allow future versions
 of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> to use different stylesheets for the packages they install.
 </p>
 </li>
 <li><p> A top-level file &lsquo;<span class="file">.Rinstignore</span>&rsquo; in the package sources
 can list (in the same way as &lsquo;<span class="file">.Rbuildignore</span>&rsquo;) files under
 <code>inst</code> that should not be installed.  (Why should there be any
 such files?  Because all the files needed to re-build vignettes
 need to be under <code>inst/doc</code>, but they may not need to be
 installed.)
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> has a new option
 <span class="option">--compact-docs</span> to compact any PDFs under the
 &lsquo;<span class="file">inst/doc</span>&rsquo; directory.  Currently this uses <code>qpdf</code>,
 which must be installed (see &lsquo;Writing R Extensions&rsquo;).
 </p>
 </li>
 <li><p> There is a new option <span class="option">--lock</span> which can be used to
 cancel the effect of <span class="option">--no-lock</span> or <span class="option">--pkglock</span> earlier
 on the command line.
 </p>
 </li>
 <li><p> Option <span class="option">--pkglock</span> can now be used with more than one
 package, and is now the default if only one package is specified.
 </p>
 </li>
 <li><p> Argument <code>lock</code> of <code>install.packages()</code> can now be
 use for Mac binary installs as well as for Windows ones.  The
 value <code>"pkglock"</code> is now accepted, as well as <code>TRUE</code> and
 <code>FALSE</code> (the default).
 </p>
 </li>
 <li><p> There is a new option <span class="option">--no-clean-on-error</span> for
 <code>R CMD INSTALL</code> to retain a partially installed package for
 forensic analysis.
 </p>
 </li>
 <li><p> Packages with names ending in &lsquo;<span class="file">.</span>&rsquo; are not portable
 since Windows does not work correctly with such directory names.
 This is now warned about in <code>R CMD check</code>, and will not be
 allowed in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.14.x.
 </p>
 </li>
 <li><p> The vignette indices are more comprehensive (in the style of
 <code>browseVignetttes()</code>).
 </p>
 </li></ul>


 <h4>DEPRECATED &amp; DEFUNCT</h4>


 <ul>
 <li> <p><code>require(save = TRUE)</code> is defunct, and use of the
 <code>save</code> argument is deprecated.
 </p>
 </li>
 <li> <p><code>R CMD check --no-latex</code> is defunct: use
 <span class="option">--no-manual</span> instead.
 </p>
 </li>
 <li> <p><code>R CMD Sd2Rd</code> is defunct.
 </p>
 </li>
 <li><p> The <code>gamma</code> argument to <code>hsv()</code>,
 <code>rainbow()</code>, and <code>rgb2hsv()</code> is deprecated
 and no longer has any effect.
 </p>
 </li>
 <li><p> The previous options for <code>R CMD build --binary</code>
 (<span class="option">--auto-zip</span>, <span class="option">--use-zip-data</span> and
 <span class="option">--no-docs</span>) are deprecated (or defunct):
 use the new option <span class="option">--install-args</span> instead.
 </p>
 </li>
 <li><p> When a character value is used for the <code>EXPR</code>
 argument in <code>switch()</code>, only a single unnamed alternative
 value is now allowed.
 </p>
 </li>
 <li><p> The wrapper <code>utils::link.html.help()</code> is no longer
 available.
 </p>
 </li>
 <li><p> Zip-ing data sets in packages (and hence <code>R CMD
 	INSTALL</code> options <span class="option">--use-zip-data</span> and <span class="option">--auto-zip</span>, as
 well as the <span class="samp">ZipData: yes</span> field in a <code>DESCRIPTION</code> file)
 is defunct.
 </p>
 <p>Installed packages with zip-ed data sets can still be used, but a
 warning that they should be re-installed will be given.
 </p>
 </li>
 <li><p> The &lsquo;experimental&rsquo; alternative specification of a
 namespace via <code>.Export()</code> etc is now defunct.
 </p>
 </li>
 <li><p> The option <span class="option">--unsafe</span> to <code>R CMD INSTALL</code> is
 deprecated: use the identical option <span class="option">--no-lock</span> instead.
 </p>
 </li>
 <li><p> The entry point <code>pythag</code> in &lsquo;<span class="file">Rmath.h</span>&rsquo; is deprecated
 in favour of the C99 function <code>hypot</code>.  A wrapper for
 <code>hypot</code> is provided for <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.x only.
 </p>
 </li>
 <li><p> Direct access to the <code>"source"</code> attribute of functions
 is deprecated; use <code>deparse(fn, control="useSource")</code> to access
 it, and <code>removeSource(fn)</code> to remove it.
 </p>
 </li>
 <li> <p><code>R CMD build --binary</code> is now formally deprecated:
 <code>R CMD INSTALL --build</code> has long been the preferred
 alternative.
 </p>
 </li>
 <li><p> Single-character package names are deprecated (and <span class="pkg">R</span>
 is already disallowed to avoid confusion in <span class="samp">Depends:</span>
 fields).
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>drop.terms</code> and the <code>[</code> method for class
 <code>"terms"</code> no longer add back an intercept.  (Reported by
 Niels Hansen.)
 </p>
 </li>
 <li> <p><code>aggregate</code> preserves the class of a column (e.g., a
 date) under some circumstances where it discarded the class
 previously.
 </p>
 </li>
 <li> <p><code>p.adjust()</code> now always returns a vector result, as
 documented.  In previous versions it copied attributes (such as
 dimensions) from the <code>p</code> argument: now it only copies names.
 </p>
 </li>
 <li><p> On PDF and PostScript devices, a line width of zero was
 recorded verbatim and this caused problems for some viewers
 (a very thin line combined with a non-solid line dash pattern
 could also cause a problem).  On these devices, the line width is
 now limited at 0.01 and for very thin lines with complex dash
 patterns the device may force the line dash pattern to be solid.
 (Reported by Jari Oksanen.)
 </p>
 </li>
 <li><p> The <code>str()</code> method for class <code>"POSIXt"</code> now gives
 sensible output for 0-length input.
 </p>
 </li>
 <li><p> The one- and two-argument complex maths functions failed to
 warn if NAs were generated (as their numeric analogues do).
 </p>
 </li>
 <li><p> Added <code>.requireCachedGenerics</code> to the <code>dont.mind</code>
 list for <code>library()</code> to avoid warnings about duplicates.
 </p>
 </li>
 <li> <p><code>$&lt;-.data.frame</code> messed with the class attribute,
 breaking any S4 subclass.  The S4 <code>data.frame</code> class now has its own
 <code>$&lt;-</code> method, and turns dispatch on for this primitive.
 </p>
 </li>
 <li> <p><code>Map()</code> did not look up a character argument <code>f</code>
 in the correct frame, thanks to lazy evaluation.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14495">PR#14495</a>)
 </p>
 </li>
 <li> <p><code>file.copy()</code> did not tilde-expand <code>from</code> and <code>to</code>
 when <code>to</code> was a directory.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14507">PR#14507</a>)
 </p>
 </li>
 <li><p> It was possible (but very rare) for the loading test in
 <code>R CMD INSTALL</code> to crash a child <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> process and so leave
 around a lock directory and a partially installed package.  That
 test is now done in a separate process.
 </p>
 </li>
 <li> <p><code>plot(&lt;formula&gt;, data=&lt;matrix&gt;,..)</code> now works in more
 cases; similarly for <code>points()</code>, <code>lines()</code> and
 <code>text()</code>.
 </p>
 </li>
 <li> <p><code>edit.default()</code> contained a manual dispatch for matrices
 (the <code>"matrix"</code> class didn't really exist when it was written).
 This caused an infinite recursion in the no-GUI case and has now
 been removed.
 </p>
 </li>
 <li> <p><code>data.frame(check.rows = TRUE)</code> sometimes worked when
 it should have detected an error.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14530">PR#14530</a>)
 </p>
 </li>
 <li> <p><code>scan(sep= , strip.white=TRUE)</code> sometimes stripped
 trailing spaces from within quoted strings.  (The real bug in
 <a href="https://bugs.R-project.org/show_bug.cgi?id=14522">PR#14522</a>.)
 </p>
 </li>
 <li><p> The rank-correlation methods for <code>cor()</code> and
 <code>cov()</code> with <code>use = "complete.obs"</code> computed the ranks
 before removing missing values, whereas the documentation implied
 incomplete cases were removed first.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14488">PR#14488</a>)
 </p>
 <p>They also failed for 1-row matrices.
 </p>
 </li>
 <li><p> The perpendicular adjustment used in placing text and
 expressions in the margins of plots was not scaled by
 <code>par("mex")</code>. (Part of <a href="https://bugs.R-project.org/show_bug.cgi?id=14532">PR#14532</a>.)
 </p>
 </li>
 <li><p> Quartz Cocoa device now catches any Cocoa exceptions that
 occur during the creation of the device window to prevent crashes.
 It also imposes a limit of 144 ft^2 on the area used by a window
 to catch user errors (unit misinterpretation) early.
 </p>
 </li>
 <li><p> The browser (invoked by <code>debug()</code>, <code>browser()</code>
 or otherwise) would display attributes such as <code>"wholeSrcref"</code>
 that were intended for internal use only.
 </p>
 </li>
 <li><p> R's internal filename completion now properly handles
 filenames with spaces in them even when the readline library is
 used.  This resolves <a href="https://bugs.R-project.org/show_bug.cgi?id=14452">PR#14452</a> provided the internal filename
 completion is used (e.g., by setting <code>rc.settings(files =
 	TRUE)</code>).
 </p>
 </li>
 <li><p> Inside <code>uniroot(f, ...)</code>, <code>-Inf</code> function values
 are now replaced by a maximally <b>negative</b> value.
 </p>
 </li>
 <li> <p><code>rowsum()</code> could silently over/underflow on integer
 inputs (reported by Bill Dunlap).
 </p>
 </li>
 <li> <p><code>as.matrix()</code> did not handle <code>"dist"</code> objects with
 zero rows.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.12.2 patched</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>max()</code> and <code>min()</code> work harder to ensure that
 <code>NA</code> has precedence over <code>NaN</code>,
 so e.g. <code>min(NaN, NA)</code> is <code>NA</code>.
 (This was not previously documented except
 for within a single numeric vector, where compiler optimizations
 often defeated the code.)
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li><p> A change to the C function <span class="samp">R_tryEval</span> had broken error
 messages in S4 method selection; the error message is now printed.
 </p>
 </li>
 <li><p> PDF output with a non-RGB color model used RGB for the line
 stroke color.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14511">PR#14511</a>)
 </p>
 </li>
 <li> <p><code>stats4::BIC()</code> assumed without checking that an object
 of class <code>"logLik"</code> has an <code>"nobs"</code> attribute:
 <code>glm()</code> fits did not and so <code>BIC()</code> failed for them.
 </p>
 </li>
 <li><p> In some circumstances a one-sided <code>mantelhaen.test()</code>
 reported the p-value for the wrong tail.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14514">PR#14514</a>)
 </p>
 </li>
 <li><p> Passing the invalid value <code>lty = NULL</code> to <code>axis()</code>
 sent an invalid value to the graphics device, and might cause the
 device to segfault.
 </p>
 </li>
 <li> <p><code>Sweave()</code> with <code>concordance=TRUE</code> could lead to
 invalid PDF files; &lsquo;<span class="file">Sweave.sty</span>&rsquo; has been updated to avoid
 this.
 </p>
 </li>
 <li><p> Non-ASCII characters in the titles of help pages were
 not rendered properly in some locales, and could cause errors
 or warnings.
 </p>
 </li>
 <li> <p><code>checkRd()</code> gave a spurious error if the <code>\href</code>
 macro was used.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.12.2</h3>


 <h4>SIGNIFICANT USER-VISIBLE CHANGES</h4>


 <ul>
 <li><p> Complex arithmetic (notably <code>z^n</code> for complex <code>z</code>
 and integer <code>n</code>) gave incorrect results since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.10.0 on
 platforms without C99 complex support.  This and some
 lesser issues in trigonometric functions have been corrected.
 </p>
 <p>Such platforms were rare (we know of Cygwin and FreeBSD).
 However, because of new compiler optimizations in the way complex
 arguments are handled, the same code was selected on x86_64 Linux
 with <code>gcc 4.5.x</code> at the default <code>-O2</code> optimization
 (but not at <code>-O</code>).
 </p>
 </li>
 <li><p> There is a workaround for crashes seen with several packages
 on systems using <span class="samp">zlib 1.2.5</span>: see the INSTALLATION section.
 </p>
 </li></ul>


 <h4>NEW FEATURES</h4>


 <ul>
 <li><p> PCRE has been updated to 8.12 (two bug-fix releases since 8.10).
 </p>
 </li>
 <li> <p><code>rep()</code>, <code>seq()</code>, <code>seq.int()</code> and
 <code>seq_len()</code> report more often when the first element is taken
 of an argument of incorrect length.
 </p>
 </li>
 <li><p> The Cocoa back-end for the <code>quartz()</code> graphics device
 on Mac OS X provides a way to disable event loop processing
 temporarily (useful, e.g., for forked instances of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>).
 </p>
 </li>
 <li> <p><code>kernel()</code>'s default for <code>m</code> was not appropriate
 if <code>coef</code> was a set of coefficients.  (Reported by Pierre
 Chausse.)
 </p>
 </li>
 <li> <p><code>bug.report()</code> has been updated for the current <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> bug
 tracker, which does not accept emailed submissions.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now checks for the correct use of
 <span class="samp">$(LAPACK_LIBS)</span> (as well as <span class="samp">$(BLAS_LIBS)</span>), since
 several CRAN recent submissions have ignored &lsquo;Writing R
 Extensions&rsquo;.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> The <span class="samp">zlib</span> sources in the distribution are now built
 with all symbols remapped: this is intended to avoid problems seen
 with packages such as <a href="https://CRAN.R-project.org/package=XML"><span class="pkg">XML</span></a> and <a href="https://CRAN.R-project.org/package=rggobi"><span class="pkg">rggobi</span></a> which link to
 &lsquo;<span class="file">zlib.so.1</span>&rsquo; on systems using <span class="samp">zlib 1.2.5</span>.
 </p>
 </li>
 <li><p> The default for <span class="env">FFLAGS</span> and <span class="env">FCFLAGS</span> with
 <code>gfortran</code> on x86_64 Linux has been changed back to <span class="samp">-g
 	-O2</span>:  however, setting <span class="samp">-g -O</span> may still be needed for
 <code>gfortran 4.3.x</code>.
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li><p> A <span class="samp">LazyDataCompression</span> field in the &lsquo;<span class="file">DESCRIPTION</span>&rsquo;
 file will be used to set the value for the
 <span class="option">--data-compress</span> option of <code>R CMD INSTALL</code>.
 </p>
 </li>
 <li><p> Files &lsquo;<span class="file">R/sysdata.rda</span>&rsquo; of more than 1Mb are now stored
 in the lazyload database using <code>xz</code> compression: this for
 example halves the installed size of package <a href="https://CRAN.R-project.org/package=Imap"><span class="pkg">Imap</span></a>.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> now ensures that directories
 installed from &lsquo;<span class="file">inst</span>&rsquo; have search permission for everyone.
 </p>
 <p>It no longer installs files &lsquo;<span class="file">inst/doc/Rplots.ps</span>&rsquo; and
 &lsquo;<span class="file">inst/doc/Rplots.pdf</span>&rsquo;.  These are almost certainly left-overs
 from <code>Sweave</code> runs, and are often large.
 </p>
 </li></ul>


 <h4>DEPRECATED &amp; DEFUNCT</h4>


 <ul>
 <li><p> The &lsquo;experimental&rsquo; alternative specification of a
 namespace via <code>.Export()</code> etc is now deprecated.
 </p>
 </li>
 <li> <p><code>zip.file.extract()</code> is now deprecated.
 </p>
 </li>
 <li><p> Zip-ing data sets in packages (and hence <code>R CMD
 	INSTALL --use-zip-data</code> and the <span class="samp">ZipData: yes</span> field in a
 <code>DESCRIPTION</code> file) is deprecated: using efficiently
 compressed &lsquo;<span class="file">.rda</span>&rsquo; images and lazy-loading of data has
 superseded it.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>identical()</code> could in rare cases generate a warning
 about non-pairlist attributes on CHARSXPs.  As these are used for
 internal purposes, the attribute check should be skipped.
 (Reported by Niels Richard Hansen).
 </p>
 </li>
 <li><p> If the filename extension (usually &lsquo;<span class="file">.Rnw</span>&rsquo;) was not
 included in a call to <code>Sweave()</code>, source references would not
 work properly and the <code>keep.source</code> option failed.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14459">PR#14459</a>)
 </p>
 </li>
 <li> <p><code>format.data.frame()</code> now keeps zero character column names.
 </p>
 </li>
 <li> <p><code>pretty(x)</code> no longer raises an error when <code>x</code>
 contains solely non-finite values. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14468">PR#14468</a>)
 </p>
 </li>
 <li><p> The <code>plot.TukeyHSD()</code> function now uses a line width of
 0.5 for its reference lines rather than <span class="samp">lwd = 0</span> (which
 caused problems for some PDF and PostScript viewers).
 </p>
 </li>
 <li><p> The <code>big.mark</code> argument to <code>prettyNum()</code>,
 <code>format()</code>, etc. was inserted reversed if it was more than
 one character long.
 </p>
 </li>
 <li> <p><code>R CMD check</code> failed to check the filenames under
 &lsquo;<span class="file">man</span>&rsquo; for Windows' reserved names.
 </p>
 </li>
 <li><p> The <code>"Date"</code> and <code>"POSIXt"</code> methods for
 <code>seq()</code> could overshoot when <code>to</code> was supplied and
 <code>by</code> was specified in months or years.
 </p>
 </li>
 <li><p> The internal method of <code>untar()</code> now restores hard
 links as file copies rather than symbolic links (which did not
 work for cross-directory links).
 </p>
 </li>
 <li> <p><code>unzip()</code> did not handle zip files which contained
 filepaths with two or more leading directories which were not in
 the zipfile and did not already exist.  (It is unclear if such
 zipfiles are valid and the third-party C code used did not support
 them, but <a href="https://bugs.R-project.org/show_bug.cgi?id=14462">PR#14462</a> created one.)
 </p>
 </li>
 <li> <p><code>combn(n, m)</code> now behaves more regularly for the border
 case <i>m = 0</i>. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14473">PR#14473</a>)
 </p>
 </li>
 <li><p> The rendering of numbers in plotmath expressions (e.g.,
 <code>expression(10^2)</code>) used the current settings for conversion
 to strings rather than setting the defaults, and so could be
 affected by what has been done before.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14477">PR#14477</a>)
 </p>
 </li>
 <li><p> The methods of <code>napredict()</code> and <code>naresid()</code> for
 <code>na.action = na.exclude</code> fits did not work correctly in the
 very rare event that every case had been omitted in the fit.
 (Reported by Simon Wood.)
 </p>
 </li>
 <li> <p><code>weighted.residuals(drop0=TRUE)</code> returned a vector when
 the residuals were a matrix (e.g., those of class <code>"mlm"</code>).
 (Reported by Bill Dunlap.)
 </p>
 </li>
 <li><p> Package HTML index files &lsquo;<span class="file">&lt;pkg&gt;/html/00Index.html</span>&rsquo; were
 generated with a stylesheet reference that was not correct for
 static browsing in libraries.
 </p>
 </li>
 <li> <p><code>ccf(na.action = na.pass)</code> was not implemented.
 </p>
 </li>
 <li><p> The parser accepted some incorrect numeric constants,
 e.g. <code>20x2</code>.  (Reported by Olaf Mersmann.)
 </p>
 </li>
 <li> <p><code>format(*, zero.print)</code> did not always replace the full
 zero parts.
 </p>
 </li>
 <li><p> Fixes for subsetting or subassignment of <code>"raster"</code>
 objects when not both <code>i</code> and <code>j</code> are specified.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> was not always respecting the
 <span class="samp">ZipData: yes</span> field of a &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file (although
 this is frequently incorrectly specified for packages with no data
 or which specify lazy-loading of data).
 </p>
 <p><code>R CMD INSTALL --use-zip-data</code> was incorrectly implemented
 as <span class="option">--use-zipdata</span> since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.9.0.
 </p>
 </li>
 <li> <p><code>source(file, echo=TRUE)</code> could fail if the file
 contained <span class="samp">#line</span> directives.  It now recovers more gracefully, but
 may still display the wrong line if the directive gives incorrect
 information.
 </p>
 </li>
 <li> <p><code>atan(1i)</code> returned <code>NaN+Infi</code> (rather than
 <code>0+Infi</code>) on platforms without C99 complex support.
 </p>
 </li>
 <li> <p><code>library()</code> failed to cache S4 metadata (unlike
 <code>loadNamespace()</code>) causing failures in S4-using packages
 without a namespace (e.g., those using reference classes).
 </p>
 </li>
 <li><p> The function <code>qlogis(lp, log.p=TRUE)</code> no longer
 prematurely overflows to <code>Inf</code> when exp(lp) is close to 1.
 </p>
 </li>
 <li><p> Updating S4 methods for a group generic function requires
 resetting the methods tables for the members of the group (patch
 contributed by Martin Morgan).
 </p>
 </li>
 <li><p> In some circumstances (including for package <a href="https://CRAN.R-project.org/package=XML"><span class="pkg">XML</span></a>),
 <code>R CMD INSTALL</code> installed version-control directories from
 source packages.
 </p>
 </li>
 <li><p> Added <code>PROTECT</code> calls to some constructed expressions
 used in C level <code>eval</code> calls.
 </p>
 </li>
 <li> <p><code>utils:::create.post()</code> (used by <code>bug.report()</code>
 and <code>help.request()</code>) failed to quote arguments to the
 mailer, and so often failed.
 </p>
 </li>
 <li> <p><code>bug.report()</code> was naive about how to extract
 maintainer email addresses from package descriptions, so would
 often try mailing to incorrect addresses.
 </p>
 </li>
 <li> <p><code>debugger()</code> could fail to read the environment of a
 call to a function with a <code>...</code> argument.  (Reported by
 Charlie Roosen.)
 </p>
 </li>
 <li> <p><code>prettyNum(c(1i, NA), drop0=TRUE)</code> or
 <code>str(NA_complex_)</code> now work correctly.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.12.1</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li><p> The DVI/PDF reference manual now includes the help pages for
 all the standard packages: <span class="pkg">splines</span>, <span class="pkg">stats4</span> and
 <span class="pkg">tcltk</span> were previously omitted (intentionally).
 </p>
 </li>
 <li> <p><a href="http://www.rforge.net">http://www.rforge.net</a> has been added to the default
 set of repositories known to <code>setRepositories()</code>.
 </p>
 </li>
 <li><p> xz-utils has been updated to version 5.0.0.
 </p>
 </li>
 <li> <p><code>reshape()</code> now makes use of <code>sep</code> when forming
 names during reshaping to wide format.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14435">PR#14435</a>)
 </p>
 </li>
 <li> <p><code>legend()</code> allows the length of lines to be set by the
 end user <em>via</em> the new argument <code>seg.len</code>.
 </p>
 </li>
 <li><p> New S4 reference class utility methods <code>copy()</code>,
 <code>field()</code>, <code>getRefClass()</code> and <code>getClass()</code> have
 been added to package <span class="pkg">methods</span>.
 </p>
 </li>
 <li><p> When a character value is used for the <code>EXPR</code>
 argument in <code>switch()</code>, a warning is given if more than
 one unnamed alternative value is given.  This will become an
 error in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.0.
 </p>
 </li>
 <li> <p><code>StructTS(type = "BSM")</code> now allows series with just
 two seasons.  (Reported by Birgit Erni.)
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> The PDF reference manual is now built as PDF version 1.5
 with object compression, which on platforms for which this is not
 the default (notably MiKTeX) halves its size.
 </p>
 </li>
 <li><p> Variable <span class="env">FCLIBS</span> can be set during configuration, for
 any additional library flags needed when linking a shared object
 with the Fortran 9x compiler.  (Needed with Solaris Studio 12.2.)
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>seq.int()</code> no longer sometimes evaluates arguments
 twice.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14388">PR#14388</a>)
 </p>
 </li>
 <li><p> The <code>data.frame</code> method of <code>format()</code> failed if a
 column name was longer than 256 bytes (the maximum length allowed
 for an <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> name).
 </p>
 </li>
 <li> <p><code>predict(&lt;lm object&gt;, type ="terms", ...)</code> failed if
 both <code>terms</code> and <code>interval</code> were specified.  (Reported
 by Bill Dunlap.)
 </p>
 <p>Also, if <code>se.fit = TRUE</code> the standard errors were reported
 for all terms, not just those selected by a non-null <code>terms</code>.
 </p>
 </li>
 <li><p> The TRE regular expressions engine could terminate <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> rather
 than give an error when given certain invalid regular expressions.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14398">PR#14398</a>)
 </p>
 </li>
 <li> <p><code>cmdscale(eig = TRUE)</code> was documented to return <i>n-1</i>
 eigenvalues but in fact only returned <code>k</code>.  It now returns
 all <i>n</i> eigenvalues.
 </p>
 <p><code>cmdscale(add = TRUE)</code> failed to centre the return
 configuration and sometimes lost the labels on the points.  Its
 return value was described wrongly (it is always a list and
 contains component <code>ac</code>).
 </p>
 </li>
 <li> <p><code>promptClass()</code> in package <span class="pkg">methods</span> now works for
 reference classes and gives a suitably specialized skeleton of
 documentation.
 </p>
 <p>Also, <code>callSuper()</code> now works via the <code>methods()</code>
 invocation as well as for initially specified methods.
 </p>
 </li>
 <li> <p><code>download.file()</code> could leave the destination file open
 if the URL was not able to be opened.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14414">PR#14414</a>)
 </p>
 </li>
 <li><p> Assignment of an environment to functions or as an attribute
 to other objects now works for S4 subclasses of <code>"environment"</code>.
 </p>
 </li>
 <li><p> Use of <span class="samp">[[&lt;-</span> for S4 subclasses of
 <code>"environment"</code> generated an infinite recursion from the
 method.  The method has been replaced by internal code.
 </p>
 </li>
 <li><p> In a reference class S4 method, <code>callSuper()</code> now works in
 <code>initialize()</code> methods when there is no explicit superclass
 method.
 </p>
 </li>
 <li> <p><span class="samp">!</span> dropped attributes such as names and dimensions from a
 length-zero argument.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14424">PR#14424</a>)
 </p>
 </li>
 <li><p> When <code>list2env()</code> created an environment it was missing
 a <code>PROTECT</code> call and so was vulnerable to garbage collection.
 </p>
 </li>
 <li> <p><code>Sweave()</code> with <code>keep.source=TRUE</code> dropped
 comments at the start and end of code chunks.  It could also
 fail when <span class="samp">\SweaveInput</span> was combined with named chunks.
 </p>
 </li>
 <li><p> The Fortran code used by <code>nls(algorithm = "port")</code>
 could infinite-loop when compiled with high optimization on a
 modern version of <code>gcc</code>, and <code>SAFE_FFLAGS</code> is now used
 to make this less likely.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14427">PR#14427</a>, seen with 32-bit Windows
 using <code>gcc 4.5.0</code> used from <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.12.0.)
 </p>
 </li>
 <li> <p><code>sapply()</code> with default <code>simplify = TRUE</code> and
 <code>mapply()</code> with default <code>SIMPLIFY = TRUE</code> wrongly
 simplified language-like results, as, e.g., in
 <code>mapply(1:2, c(3,7), FUN = function(i,j) call(':',i,j))</code>.
 </p>
 </li>
 <li><p> Backreferences to undefined patterns in
 <code>[g]sub(pcre = TRUE)</code> could cause a segfault.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14431">PR#14431</a>)
 </p>
 </li>
 <li><p> The <code>format()</code> (and hence the <code>print()</code>) method
 for class <code>"Date"</code> rounded fractional dates towards zero: it
 now always rounds them down.
 </p>
 </li>
 <li><p> Reference S4 class creation could generate ambiguous
 inheritance patterns under very special circumstances.
 </p>
 </li>
 <li> <p><span class="samp">[[&lt;-</span> turned S4 subclasses of <code>"environment"</code> into
 plain environments.
 </p>
 </li>
 <li><p> Long titles for help pages were truncated in package indices
 and a few other places.
 </p>
 </li>
 <li><p> Additional utilities now work correctly with S4 subclasses of
 <code>"environment"</code> (<code>rm</code>, locking tools and active
 bindings).
 </p>
 </li>
 <li> <p><code>spec.ar()</code> now also work for the <code>"ols"</code> method.
 (Reported by Hans-Ruedi Kuensch.)
 </p>
 </li>
 <li><p> The initialization of objects from S4 subclasses of
 <code>"environment"</code> now allocates a new environment object.
 </p>
 </li>
 <li> <p><code>R CMD check</code> has more protection against (probably
 erroneous) example or test output which is invalid in the current
 locale.
 </p>
 </li>
 <li> <p><code>qr.X()</code> with column names and pivoting now also pivots
 the column names.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14438">PR#14438</a>)
 </p>
 </li>
 <li> <p><code>unit.pmax()</code> and <code>unit.pmin()</code> in package
 <span class="pkg">grid</span> gave incorrect results when all inputs were of length 1.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14443">PR#14443</a>)
 </p>
 </li>
 <li><p> The parser for &lsquo;<span class="file">NAMESPACE</span>&rsquo; files ignored misspelled
 directives, rather than signalling an error.  For 2.12.x a warning
 will be issued, but this will be correctly reported as an error
 in later releases. (Reported by Charles Berry.)
 </p>
 </li>
 <li><p> Fix for subsetting of <code>"raster"</code> objects when only one
 of <code>i</code> or <code>j</code> is specified.
 </p>
 </li>
 <li> <p><code>grid.raster()</code> in package <span class="pkg">grid</span> did not accept
 <code>"nativeRaster"</code> objects (like <code>rasterImage()</code> does).
 </p>
 </li>
 <li><p> Rendering raster images in PDF output was resetting the
 clipping region.
 </p>
 </li>
 <li><p> Rendering of raster images on cairo X11 device was wrong,
 particularly when a small image was being scaled up using
 interpolation.
 </p>
 <p>With cairo &lt; 1.6, will be better than before, though still a
 little clunky.  With cairo &gt;= 1.6, should be sweet as.
 </p>
 </li>
 <li><p> Several bugs fixed in <code>read.DIF()</code>:  single column
 inputs caused errors, cells marked as <code>"character"</code> could
 be converted to other types, and (in Windows) copying from the
 clipboard failed.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.12.0</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li><p> Reading a package's &lsquo;<span class="file">CITATION</span>&rsquo; file now defaults to
 ASCII rather than Latin-1: a package with a non-ASCII
 &lsquo;<span class="file">CITATION</span>&rsquo; file should declare an encoding in its
 &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file and use that encoding for the
 &lsquo;<span class="file">CITATION</span>&rsquo; file.
 </p>
 </li>
 <li> <p><code>difftime()</code> now defaults to the <code>"tzone"</code>
 attribute of <code>"POSIXlt"</code> objects rather than to the current
 timezone as set by the default for the <code>tz</code> argument.  (Wish
 of <a href="https://bugs.R-project.org/show_bug.cgi?id=14182">PR#14182</a>.)
 </p>
 </li>
 <li> <p><code>pretty()</code> is now generic, with new methods for
 <code>"Date"</code> and <code>"POSIXt"</code> classes (based on code
 contributed by Felix Andrews).
 </p>
 </li>
 <li> <p><code>unique()</code> and <code>match()</code> are now faster on
 character vectors where all elements are in the global <code>CHARSXP</code>
 cache and have unmarked encoding (ASCII).  Thanks to Matthew
 Dowle for suggesting improvements to the way the hash code is
 generated in &lsquo;<span class="file">unique.c</span>&rsquo;.
 </p>
 </li>
 <li><p> The <code>enquote()</code> utility, in use internally, is exported now.
 </p>
 </li>
 <li> <p><code>.C()</code> and <code>.Fortran()</code> now map non-zero return
 values (other than <code>NA_LOGICAL</code>) for logical vectors to
 <code>TRUE</code>: it has been an implicit assumption that they are
 treated as true.
 </p>
 </li>
 <li><p> The <code>print()</code> methods for <code>"glm"</code> and <code>"lm"</code>
 objects now insert linebreaks in long calls in the same way that
 the <code>print()</code> methods for <code>"summary.[g]lm"</code> objects
 have long done.  This does change the layout of the examples for
 a number of packages, e.g. <a href="https://CRAN.R-project.org/package=MASS"><span class="pkg">MASS</span></a>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14250">PR#14250</a>)
 </p>
 </li>
 <li> <p><code>constrOptim()</code> can now be used with method
 <code>"SANN"</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14245">PR#14245</a>)
 </p>
 <p>It gains an argument <code>hessian</code> to be passed to
 <code>optim()</code>, which allows all the <code>...</code> arguments to be
 intended for <code>f()</code> and <code>grad()</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14071">PR#14071</a>)
 </p>
 </li>
 <li> <p><code>curve()</code> now allows <code>expr</code> to be an object of
 mode <code>"expression"</code> as well as <code>"call"</code> and
 <code>"function"</code>.
 </p>
 </li>
 <li><p> The <code>"POSIX[cl]t"</code> methods for <code>Axis()</code> have been
 replaced by a single method for <code>"POSIXt"</code>.
 </p>
 <p>There are no longer separate <code>plot()</code> methods for
 <code>"POSIX[cl]t"</code> and <code>"Date"</code>: the default method has
 been able to handle those classes for a long time.  This
 <em>inter alia</em> allows a single date-time object to be
 supplied, the wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14016">PR#14016</a>.
 </p>
 <p>The methods had a different default (<code>""</code>) for <code>xlab</code>.
 </p>
 </li>
 <li><p> Classes <code>"POSIXct"</code>, <code>"POSIXlt"</code> and
 <code>"difftime"</code> have generators <code>.POSIXct()</code>,
 <code>.POSIXlt()</code> and <code>.difftime()</code>.  Package authors are
 advised to make use of them (they are available from <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.11.0) to
 proof against planned future changes to the classes.
 </p>
 <p>The ordering of the classes has been changed, so <code>"POSIXt"</code>
 is now the second class.  See the document &lsquo;Updating
 packages for changes in R 2.12.x&rsquo; on
 <a href="http://developer.r-project.org">http://developer.r-project.org</a> for the consequences for a
 handful of CRAN packages.
 </p>
 </li>
 <li><p> The <code>"POSIXct"</code> method of <code>as.Date()</code> allows a
 timezone to be specified (but still defaults to UTC).
 </p>
 </li>
 <li><p> New <code>list2env()</code> utility function as an inverse of
 <code>as.list(&lt;environment&gt;)</code> and for fast multi-<code>assign()</code>
 to existing environment.  <code>as.environment()</code> is now generic
 and uses <code>list2env()</code> as list method.
 </p>
 </li>
 <li><p> There are several small changes to output which &lsquo;zap&rsquo;
 small numbers, e.g. in printing quantiles of residuals in
 summaries from <code>"lm"</code> and <code>"glm"</code> fits, and in test
 statistics in <code>print.anova()</code>.
 </p>
 </li>
 <li><p> Special names such as <code>"dim"</code>, <code>"names"</code>, etc, are
 now allowed as slot names of S4 classes, with <code>"class"</code> the
 only remaining exception.
 </p>
 </li>
 <li><p> File &lsquo;<span class="file">.Renviron</span>&rsquo; can have architecture-specific versions
 such as &lsquo;<span class="file">.Renviron.i386</span>&rsquo; on systems with sub-architectures.
 </p>
 </li>
 <li> <p><code>installed.packages()</code> has a new argument
 <code>subarch</code> to filter on sub-architecture.
 </p>
 </li>
 <li><p> The <code>summary()</code> method for <code>packageStatus()</code> now
 has a separate <code>print()</code> method.
 </p>
 </li>
 <li><p> The default <code>summary()</code> method returns an object
 inheriting from class <code>"summaryDefault"</code> which has a
 separate <code>print()</code> method that calls <code>zapsmall()</code> for
 numeric/complex values.
 </p>
 </li>
 <li><p> The startup message now includes the platform and if used,
 sub-architecture: this is useful where different
 (sub-)architectures run on the same OS.
 </p>
 </li>
 <li><p> The <code>getGraphicsEvent()</code> mechanism now allows multiple windows
 to return graphics events, through the new functions
 <code>setGraphicsEventHandlers()</code>, <code>setGraphicsEventEnv()</code>, and
 <code>getGraphicsEventEnv()</code>.  (Currently implemented in the
 <code>windows()</code> and <code>X11()</code> devices.)
 </p>
 </li>
 <li> <p><code>tools::texi2dvi()</code> gains an <code>index</code> argument,
 mainly for use by <code>R CMD Rd2pdf</code>.
 </p>
 <p>It avoids the use of <code>texindy</code> by <code>texinfo</code>'s
 <code>texi2dvi &gt;= 1.157</code>, since that does not emulate
 'makeindex' well enough to avoid problems with special characters
 (such as <span class="samp">(</span>, <span class="samp">{</span>, <span class="samp">!</span>) in indices.
 </p>
 </li>
 <li><p> The ability of <code>readLines()</code> and <code>scan()</code> to
 re-encode inputs to marked UTF-8 strings on Windows since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.7.0
 is extended to non-UTF-8 locales on other OSes.
 </p>
 </li>
 <li> <p><code>scan()</code> gains a <code>fileEncoding</code> argument to
 match <code>read.table()</code>.
 </p>
 </li>
 <li> <p><code>points()</code> and <code>lines()</code> gain <code>"table"</code>
 methods to match <code>plot()</code>.  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=10472">PR#10472</a>.)
 </p>
 </li>
 <li> <p><code>Sys.chmod()</code> allows argument <code>mode</code> to be a
 vector, recycled along <code>paths</code>.
 </p>
 </li>
 <li><p> There are <code>|</code>, <code>&amp;</code> and <code>xor()</code> methods for
 classes <code>"octmode"</code> and <code>"hexmode"</code>, which work
 bitwise.
 </p>
 </li>
 <li><p> Environment variables <span class="env">R_DVIPSCMD</span>, <span class="env">R_LATEXCMD</span>,
 <span class="env">R_MAKEINDEXCMD</span>, <span class="env">R_PDFLATEXCMD</span> are no longer used
 nor set in an <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> session.  (With the move to
 <code>tools::texi2dvi()</code>, the conventional environment variables
 <span class="env">LATEX</span>, <span class="env">MAKEINDEX</span> and <span class="env">PDFLATEX</span> will be used.
 <code>options("dvipscmd")</code> defaults to the value of <span class="env">DVIPS</span>,
 then to <code>"dvips"</code>.)
 </p>
 </li>
 <li><p> New function <code>isatty()</code> to see if terminal connections are
 redirected.
 </p>
 </li>
 <li> <p><code>summaryRprof()</code> returns the sampling interval in
 component <code>sample.interval</code> and only returns in
 <code>by.self</code> data for functions with non-zero self times.
 </p>
 </li>
 <li> <p><code>print(x)</code> and <code>str(x)</code> now indicate if an empty
 list <code>x</code> is named.
 </p>
 </li>
 <li> <p><code>install.packages()</code> and <code>remove.packages()</code> with
 <code>lib</code> unspecified and multiple libraries in
 <code>.libPaths()</code> inform the user of the library location used
 with a message rather than a warning.
 </p>
 </li>
 <li><p> There is limited support for multiple compressed streams on a
 file: all of <code>[bgx]zfile()</code> allow streams to be appended to an
 existing file, but <code>bzfile()</code> reads only the first stream.
 </p>
 </li>
 <li><p> Function <code>person()</code> in package <span class="pkg">utils</span> now uses
 a given/family scheme in preference to first/middle/last, is
 vectorized to handle an arbitrary number of persons, and gains a
 <code>role</code> argument to specify person roles using a controlled
 vocabulary (the MARC relator terms).
 </p>
 </li>
 <li><p> Package <span class="pkg">utils</span> adds a new <code>"bibentry"</code> class
 for representing and manipulating bibliographic information in
 enhanced BibTeX style, unifying and enhancing the previously
 existing mechanisms.
 </p>
 </li>
 <li><p> A <code>bibstyle()</code> function has been added to the
 <span class="pkg">tools</span> package with default JSS style for rendering
 <code>"bibentry"</code> objects, and a mechanism for registering other
 rendering styles.
 </p>
 </li>
 <li><p> Several aspects of the display of text help are now
 customizable using the new <code>Rd2txt_options()</code> function.
 options(<code>"help_text_width"</code>) is no longer used.
 </p>
 </li>
 <li><p> Added <span class="samp">\href</span> tag to the Rd format, to allow
 hyperlinks to URLs without displaying the full URL.
 </p>
 </li>
 <li><p> Added <span class="samp">\newcommand</span> and <span class="samp">\renewcommand</span> tags
 to the Rd format, to allow user-defined macros.
 </p>
 </li>
 <li><p> New <code>toRd()</code> generic in the <span class="pkg">tools</span> package to
 convert objects to fragments of Rd code, and added
 <code>"fragment"</code> argument to <code>Rd2txt()</code>, <code>Rd2HTML()</code>,
 and <code>Rd2latex()</code> to support it.
 </p>
 </li>
 <li><p> Directory &lsquo;<span class="file">R_HOME/share/texmf</span>&rsquo; now follows the TDS
 conventions, so can be set as a <code>texmf</code> tree (&lsquo;root
 directory&rsquo; in MiKTeX parlance).
 </p>
 </li>
 <li><p> S3 generic functions now use correct S4 inheritance when
 dispatching on an S4 object.  See <code>?Methods</code>, section on
 &ldquo;Methods for S3 Generic Functions&rdquo; for recommendations
 and details.
 </p>
 </li>
 <li> <p><code>format.pval()</code> gains a <code>...</code> argument to pass
 arguments such as <code>nsmall</code> to <code>format()</code>.  (Wish of
 <a href="https://bugs.R-project.org/show_bug.cgi?id=9574">PR#9574</a>)
 </p>
 </li>
 <li> <p><code>legend()</code> supports <code>title.adj</code>.  (Wish of
 <a href="https://bugs.R-project.org/show_bug.cgi?id=13415">PR#13415</a>)
 </p>
 </li>
 <li><p> Added support for subsetting <code>"raster"</code> objects, plus
 assigning to a subset, conversion to a matrix (of colour
 strings), and comparisons (<span class="samp">==</span> and <span class="samp">!=</span>).
 </p>
 </li>
 <li><p> Added a new <code>parseLatex()</code> function (and related
 functions <code>deparseLatex()</code> and <code>latexToUtf8()</code>) to
 support conversion of bibliographic entries for display in R.
 </p>
 </li>
 <li><p> Text rendering of <span class="samp">\itemize</span> in help uses a Unicode
 bullet in UTF-8 and most single-byte Windows locales.
 </p>
 </li>
 <li><p> Added support for polygons with holes to the graphics
 engine.  This is implemented for the <code>pdf()</code>,
 <code>postscript()</code>, <code>x11(type="cairo")</code>, <code>windows()</code>,
 and <code>quartz()</code> devices (and associated raster formats),
 but not for <code>x11(type="Xlib")</code> or <code>xfig()</code> or
 <code>pictex()</code>.  The user-level interface is the
 <code>polypath()</code> function in <span class="pkg">graphics</span> and
 <code>grid.path()</code> in <span class="pkg">grid</span>.
 </p>
 </li>
 <li><p> File &lsquo;<span class="file">NEWS</span>&rsquo; is now generated at installation with a
 slightly different format: it will be in UTF-8 on platforms using
 UTF-8, and otherwise in ASCII.  There is also a PDF version,
 &lsquo;<span class="file">NEWS.pdf</span>&rsquo;, installed at the top-level of the R distribution.
 </p>
 </li>
 <li> <p><code>kmeans(x, 1)</code> now works.  Further, <code>kmeans</code> now
 returns between and total sum of squares.
 </p>
 </li>
 <li> <p><code>arrayInd()</code> and <code>which()</code> gain an argument
 <code>useNames</code>.  For <code>arrayInd</code>, the default is now false,
 for speed reasons.
 </p>
 </li>
 <li><p> As is done for closures, the default <code>print</code> method for
 the formula class now displays the associated environment if it is
 not the global environment.
 </p>
 </li>
 <li><p> A new facility has been added for inserting code into a
 package without re-installing it, to facilitate testing
 changes which can be selectively added and backed out.  See
 <code>?insertSource</code>.
 </p>
 </li>
 <li><p> New function <code>readRenviron</code> to (re-)read files in the
 format of &lsquo;<span class="file">~/.Renviron</span>&rsquo; and &lsquo;<span class="file">Renviron.site</span>&rsquo;.
 </p>
 </li>
 <li> <p><code>require()</code> will now return <code>FALSE</code> (and not fail)
 if loading the package or one of its dependencies fails.
 </p>
 </li>
 <li> <p><code>aperm()</code> now allows argument <code>perm</code> to be a
 character vector when the array has named dimnames (as the results
 of <code>table()</code> calls do).  Similarly, <code>array()</code> allows
 <code>MARGIN</code> to be a character vector.  (Based on suggestions of
 Michael Lachmann.)
 </p>
 </li>
 <li><p> Package <span class="pkg">utils</span> now exports and documents functions
 <code>aspell_package_Rd_files()</code> and
 <code>aspell_package_vignettes()</code> for spell checking package Rd
 files and vignettes using Aspell, Ispell or Hunspell.
 </p>
 </li>
 <li><p> Package news can now be given in Rd format, and
 <code>news()</code> prefers these &lsquo;<span class="file">inst/NEWS.Rd</span>&rsquo; files to old-style
 plain text &lsquo;<span class="file">NEWS</span>&rsquo; or &lsquo;<span class="file">inst/NEWS</span>&rsquo; files.
 </p>
 </li>
 <li><p> New simple function <code>packageVersion()</code>.
 </p>
 </li>
 <li><p> The PCRE library has been updated to version 8.10.
 </p>
 </li>
 <li><p> The standard Unix-alike terminal interface declares its name
 to readline as 'R', so that can be used for conditional sections
 in &lsquo;<span class="file">~/.inputrc</span>&rsquo; files.
 </p>
 </li>
 <li> <p>&lsquo;Writing R Extensions&rsquo; now stresses that the standard
 sections in &lsquo;<span class="file">.Rd</span>&rsquo; files (other than <span class="samp">\alias</span>,
 <span class="samp">\keyword</span> and <span class="samp">\note</span>) are intended to be unique, and
 the conversion tools now drop duplicates with a warning.
 </p>
 <p>The &lsquo;<span class="file">.Rd</span>&rsquo; conversion tools also warn about an unrecognized type
 in a <span class="samp">\docType</span> section.
 </p>
 </li>
 <li> <p><code>ecdf()</code> objects now have a <code>quantile()</code> method.
 </p>
 </li>
 <li> <p><code>format()</code> methods for date-time objects now attempt to make
 use of a <code>"tzone"</code> attribute with <code>"%Z"</code> and
 <code>"%z"</code> formats, but it is not always possible.  (Wish of
 <a href="https://bugs.R-project.org/show_bug.cgi?id=14358">PR#14358</a>.)
 </p>
 </li>
 <li> <p><code>tools::texi2dvi(file, clean = TRUE)</code> now works in more
 cases (e.g., where emulation is used and when &lsquo;<span class="file">file</span>&rsquo; is not in
 the current directory).
 </p>
 </li>
 <li><p> New function <code>droplevels()</code> to remove unused factor levels.
 </p>
 </li>
 <li> <p><code>system(command, intern = TRUE)</code> now gives an error on
 a Unix-alike (as well as on Windows) if <code>command</code> cannot be
 run.  It reports a non-success exit status from running
 <code>command</code> as a warning.
 </p>
 <p>On a Unix-alike an attempt is made to return the actual exit
 status of the command in <code>system(intern = FALSE)</code>: previously
 this had been system-dependent but on POSIX-compliant systems the
 value return was 256 times the status.
 </p>
 </li>
 <li> <p><code>system()</code> has a new argument <code>ignore.stdout</code>
 which can be used to (portably) ignore standard output.
 </p>
 </li>
 <li> <p><code>system(intern = TRUE)</code> and <code>pipe()</code> connections
 are guaranteed to be available on all builds of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 </p>
 </li>
 <li> <p><code>Sys.which()</code> has been altered to return <code>""</code> if
 the command is not found (even on Solaris).
 </p>
 </li>
 <li><p> A facility for defining reference-based S4 classes (in the
 OOP style of Java, C++, etc.) has been added experimentally to
 package <span class="pkg">methods</span>; see <code>?ReferenceClasses</code>.
 </p>
 </li>
 <li><p> The <code>predict</code> method for <code>"loess"</code> fits gains an
 <code>na.action</code> argument which defaults to <code>na.pass</code> rather
 than the previous default of <code>na.omit</code>.
 </p>
 <p>Predictions from <code>"loess"</code> fits are now named from the row
 names of <code>newdata</code>.
 </p>
 </li>
 <li><p> Parsing errors detected during <code>Sweave()</code> processing
 will now be reported referencing their original location in the
 source file.
 </p>
 </li>
 <li><p> New <code>adjustcolor()</code> utility, e.g., for simple
 translucent color schemes.
 </p>
 </li>
 <li> <p><code>qr()</code> now has a trivial <code>lm</code> method with a
 simple (fast) validity check.
 </p>
 </li>
 <li><p> An experimental new programming model has been added to
 package <span class="pkg">methods</span> for reference (OOP-style) classes and
 methods.  See <code>?ReferenceClasses</code>.
 </p>
 </li>
 <li><p> bzip2 has been updated to version 1.0.6 (bug-fix release).
 <span class="option">--with-system-bzlib</span> now requires at least version 1.0.6.
 </p>
 </li>
 <li><p> R now provides &lsquo;<span class="file">jss.cls</span>&rsquo; and &lsquo;<span class="file">jss.bst</span>&rsquo; (the class
 and bib style file for the Journal of Statistical Software) as
 well as &lsquo;<span class="file">RJournal.bib</span>&rsquo; and &lsquo;<span class="file">Rnews.bib</span>&rsquo;, and
 <code>R CMD</code> ensures that the &lsquo;<span class="file">.bst</span>&rsquo; and &lsquo;<span class="file">.bib</span>&rsquo; files
 are found by BibTeX.
 </p>
 </li>
 <li><p> Functions using the <span class="env">TAR</span> environment variable no longer
 quote the value when making <code>system</code> calls.  This allows
 values such as <span class="samp">tar --force-local</span>, but does require
 additional quotes in, e.g., <code>TAR = "'/path with spaces/mytar'"</code>.
 </p>
 </li></ul>


 <h4>DEPRECATED &amp; DEFUNCT</h4>


 <ul>
 <li><p> Supplying the parser with a character string containing both
 octal/hex and Unicode escapes is now an error.
 </p>
 </li>
 <li><p> File extension &lsquo;<span class="file">.C</span>&rsquo; for C++ code files in packages is
 now defunct.
 </p>
 </li>
 <li> <p><code>R CMD check</code> no longer supports configuration
 files containing Perl configuration variables: use the environment
 variables documented in &lsquo;R Internals&rsquo; instead.
 </p>
 </li>
 <li><p> The <code>save</code> argument of <code>require()</code> now defaults to
 <code>FALSE</code> and <code>save = TRUE</code> is now deprecated.  (This
 facility is very rarely actually used, and was superseded by the
 <span class="samp">Depends</span> field of the &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file long ago.)
 </p>
 </li>
 <li> <p><code>R CMD check --no-latex</code> is deprecated in favour of
 <span class="option">--no-manual</span>.
 </p>
 </li>
 <li> <p><code>R CMD Sd2Rd</code> is formally deprecated and will be
 removed in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.13.0.
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li> <p><code>install.packages()</code> has a new argument
 <code>libs_only</code> to optionally pass <span class="option">--libs-only</span> to
 <code>R CMD INSTALL</code> and works analogously for Windows binary
 installs (to add support for 64- or 32-bit Windows).
 </p>
 </li>
 <li><p> When sub-architectures are in use, the installed
 architectures are recorded in the <code>Archs</code> field of the
 &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file.  There is a new default filter,
 <code>"subarch"</code>, in <code>available.packages()</code> to make use of
 this.
 </p>
 <p>Code is compiled in a copy of the &lsquo;<span class="file">src</span>&rsquo; directory when a
 package is installed for more than one sub-architecture: this
 avoid problems with cleaning the sources between building
 sub-architectures.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL --libs-only</code> no longer overrides the
 setting of locking, so a previous version of the package will be
 restored unless <span class="option">--no-lock</span> is specified.
 </p>
 </li></ul>


 <h4>UTILITIES</h4>


 <ul>
 <li> <p><code>R CMD Rprof|build|check</code> are now based on <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> rather
 than Perl scripts.  The only remaining Perl scripts are the
 deprecated <code>R CMD Sd2Rd</code> and <code>install-info.pl</code> (used
 only if <code>install-info</code> is not found) as well as some
 maintainer-mode-only scripts.
 </p>
 <p><strong>NB:</strong> because these have been completely rewritten, users
 should not expect undocumented details of previous
 implementations to have been duplicated.
 </p>
 <p><code>R CMD</code> no longer manipulates the environment variables
 <span class="env">PERL5LIB</span> and <span class="env">PERLLIB</span>.
 </p>
 </li>
 <li> <p><code>R CMD check</code> has a new argument <span class="option">--extra-arch</span>
 to confine tests to those needed to check an additional
 sub-architecture.
 </p>
 <p>Its check for &ldquo;Subdirectory 'inst' contains no files&rdquo; is
 more thorough: it looks for files, and warns if there are only
 empty directories.
 </p>
 <p>Environment variables such as <span class="env">R_LIBS</span> and those used for
 customization can be set for the duration of checking <em>via</em> a file
 &lsquo;<span class="file">~/.R/check.Renviron</span>&rsquo; (in the format used by
 &lsquo;<span class="file">.Renviron</span>&rsquo;, and with sub-architecture specific versions
 such as &lsquo;<span class="file">~/.R/check.Renviron.i386</span>&rsquo; taking precedence).
 </p>
 <p>There are new options <span class="option">--multiarch</span> to check the package
 under all of the installed sub-architectures and
 <span class="option">--no-multiarch</span> to confine checking to the
 sub-architecture under which <code>check</code> is invoked.  If
 neither option is supplied, a test is done of installed
 sub-architectures and all those which can be run on the current
 OS are used.
 </p>
 <p>Unless multiple sub-architectures are selected, the install done
 by <code>check</code> for testing purposes is only of the current
 sub-architecture (<em>via</em> <code>R CMD INSTALL --no-multiarch</code>).
 </p>
 <p>It will skip the check for non-ascii characters in code or data
 if the environment variables <span class="env">_R_CHECK_ASCII_CODE_</span> or
 <span class="env">_R_CHECK_ASCII_DATA_</span> are respectively set to <span class="env">FALSE</span>.
 (Suggestion of Vince Carey.)
 </p>
 </li>
 <li> <p><code>R CMD build</code> no longer creates an &lsquo;<span class="file">INDEX</span>&rsquo; file
 (<code>R CMD INSTALL</code> does so), and <span class="option">--force</span> removes
 (rather than overwrites) an existing &lsquo;<span class="file">INDEX</span>&rsquo; file.
 </p>
 <p>It supports a file &lsquo;<span class="file">~/.R/build.Renviron</span>&rsquo; analogously to
 <code>check</code>.
 </p>
 <p>It now runs build-time <span class="samp">\Sexpr</span> expressions in help files.
 </p>
 </li>
 <li> <p><code>R CMD Rd2dvi</code> makes use of <code>tools::texi2dvi()</code>
 to process the package manual.  It is now implemented entirely in
 <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> (rather than partially as a shell script).
 </p>
 </li>
 <li> <p><code>R CMD Rprof</code> now uses <code>utils::summaryRprof()</code>
 rather than Perl.  It has new arguments to select one of the
 tables and to limit the number of entries printed.
 </p>
 </li>
 <li> <p><code>R CMD Sweave</code> now runs <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> with <span class="option">--vanilla</span> so
 the environment setting of <span class="env">R_LIBS</span> will always be used.
 </p>
 </li></ul>


 <h4>C-LEVEL FACILITIES</h4>


 <ul>
 <li> <p><code>lang5()</code> and <code>lang6()</code> (in addition to
 pre-existing <code>lang[1-4]()</code>) convenience functions for easier
 construction of <code>eval()</code> calls.  If you have your own
 definition, do wrap it inside <code>#ifndef lang5 .... #endif</code>
 to keep it working with old and new <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 </p>
 </li>
 <li><p> Header &lsquo;<span class="file">R.h</span>&rsquo; now includes only the C headers it itself needs,
 hence no longer includes <code>errno.h</code>.  (This helps avoid problems
 when it is included from C++ source files.)
 </p>
 </li>
 <li><p> Headers &lsquo;<span class="file">Rinternals.h</span>&rsquo; and &lsquo;<span class="file">R_ext/Print.h</span>&rsquo; include
 the C++ versions of &lsquo;<span class="file">stdio.h</span>&rsquo; and &lsquo;<span class="file">stdarg.h</span>&rsquo;
 respectively if included from a C++ source file.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> A C99 compiler is now required, and more C99 language features
 will be used in the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> sources.
 </p>
 </li>
 <li><p> Tcl/Tk &gt;= 8.4 is now required (increased from 8.3).
 </p>
 </li>
 <li><p> System functions <code>access</code>, <code>chdir</code> and
 <code>getcwd</code> are now essential to configure <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.  (In practice
 they have been required for some time.)
 </p>
 </li>
 <li> <p><code>make check</code> compares the output of the examples
 from several of the base packages to reference output rather than
 the previous output (if any).  Expect some differences due to
 differences in floating-point computations between platforms.
 </p>
 </li>
 <li><p> File &lsquo;<span class="file">NEWS</span>&rsquo; is no longer in the sources, but generated as
 part of the installation.  The primary source for changes is now
 &lsquo;<span class="file">doc/NEWS.Rd</span>&rsquo;.
 </p>
 </li>
 <li><p> The <code>popen</code> system call is now required to build <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 This ensures the availability of <code>system(intern = TRUE)</code>,
 <code>pipe()</code> connections and printing from <code>postscript()</code>.
 </p>
 </li>
 <li><p> The <code>pkg-config</code> file &lsquo;<span class="file">libR.pc</span>&rsquo; now also works
 when <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> is installed using a sub-architecture.
 </p>
 </li>
 <li> <p><span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> has always required a BLAS that conforms to IE60559
 arithmetic, but after discovery of more real-world problems caused
 by a BLAS that did not, this is tested more thoroughly in this
 version.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li><p> Calls to <code>selectMethod()</code> by default no longer cache
 inherited methods.  This could previously corrupt methods used by
 <code>as()</code>.
 </p>
 </li>
 <li><p> The densities of non-central chi-squared are now more accurate
 in some cases in the extreme tails,
 e.g. <code>dchisq(2000, 2, 1000)</code>,
 as a series expansion was truncated too early.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14105">PR#14105</a>)
 </p>
 </li>
 <li> <p><code>pt()</code> is more accurate in the left tail for <code>ncp</code>
 large, e.g. <code>pt(-1000, 3, 200)</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14069">PR#14069</a>)
 </p>
 </li>
 <li><p> The default C function (<code>R_binary</code>) for binary ops now
 sets the S4 bit in the result if either argument is an S4 object.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=13209">PR#13209</a>)
 </p>
 </li>
 <li> <p><code>source(echo=TRUE)</code> failed to echo comments that
 followed the last statement in a file.
 </p>
 </li>
 <li><p> S4 classes that contained one of <code>"matrix"</code>,
 <code>"array"</code> or <code>"ts"</code> and also another class now accept
 superclass objects in <code>new()</code>.  Also fixes failure to call
 <code>validObject()</code> for these classes.
 </p>
 </li>
 <li><p> Conditional inheritance defined by argument <code>test</code> in
 <code>methods::setIs()</code> will no longer be used in S4 method
 selection (caching these methods could give incorrect results).
 See <code>?setIs</code>.
 </p>
 </li>
 <li><p> The signature of an implicit generic is now used by
 <code>setGeneric()</code> when that does not use a definition nor
 explicitly set a signature.
 </p>
 </li>
 <li><p> A bug in <code>callNextMethod()</code> for some examples with
 <code>"..."</code> in the arguments has been fixed.  See file
 &lsquo;<span class="file">src/library/methods/tests/nextWithDots.R</span>&rsquo; in the sources.
 </p>
 </li>
 <li> <p><code>match(x, table)</code> (and hence <code>%in%</code>) now treat
 <code>"POSIXlt"</code> consistently with, e.g., <code>"POSIXct"</code>.
 </p>
 </li>
 <li><p> Built-in code dealing with environments (<code>get()</code>,
 <code>assign()</code>, <code>parent.env()</code>, <code>is.environment()</code> and
 others) now behave consistently to recognize S4 subclasses;
 <code>is.name()</code> also recognizes subclasses.
 </p>
 </li>
 <li><p> The <code>abs.tol</code> control parameter to <code>nlminb()</code>
 now defaults to <code>0.0</code> to avoid false declarations of
 convergence in objective functions that may go negative.
 </p>
 </li>
 <li><p> The standard Unix-alike termination dialog to ask whether to
 save the workspace takes a EOF response as <code>n</code> to avoid
 problems with a damaged terminal connection.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14332">PR#14332</a>)
 </p>
 </li>
 <li><p> Added <code>warn.unused</code> argument to <code>hist.default()</code>
 to allow suppression of spurious warnings about graphical
 parameters used with <code>plot=FALSE</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14341">PR#14341</a>)
 </p>
 </li>
 <li> <p><code>predict.lm()</code>, <code>summary.lm()</code>, and indeed
 <code>lm()</code> itself had issues with residual DF in zero-weighted
 cases (the latter two only in connection with empty
 models). (Thanks to Bill Dunlap for spotting the
 <code>predict()</code> case.)
 </p>
 </li>
 <li> <p><code>aperm()</code> treated <code>resize = NA</code> as <code>resize =
       TRUE</code>.
 </p>
 </li>
 <li> <p><code>constrOptim()</code> now has an improved convergence
 criterion, notably for cases where the minimum was (very close to)
 zero; further, other tweaks inspired from code proposals by Ravi
 Varadhan.
 </p>
 </li>
 <li><p> Rendering of S3 and S4 methods in man pages has been
 corrected and made consistent across output formats.
 </p>
 </li>
 <li><p> Simple markup is now allowed in <span class="samp">\title</span> sections
 in &lsquo;<span class="file">.Rd</span>&rsquo; files.
 </p>
 </li>
 <li><p> The behaviour of <code>as.logical()</code> on factors (to use the
 levels) was lost in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.6.0 and has been restored.
 </p>
 </li>
 <li> <p><code>prompt()</code> did not backquote some default arguments in
 the <span class="samp">\usage</span> section.  (Reported by Claudia Beleites.)
 </p>
 </li>
 <li> <p><code>writeBin()</code> disallows attempts to write 2GB or
 more in a single call. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14362">PR#14362</a>)
 </p>
 </li>
 <li> <p><code>new()</code> and <code>getClass()</code> will now work if Class
 is a subclass of <code>"classRepresentation"</code> and should also be
 faster in typical calls.
 </p>
 </li>
 <li><p> The <code>summary()</code> method for data frames makes a better
 job of names containing characters invalid in the current locale.
 </p>
 </li>
 <li> <p><code>[[</code> sub-assignment for factors could create an invalid
 factor (reported by Bill Dunlap).
 </p>
 </li>
 <li> <p><code>Negate(f)</code> would not evaluate argument <code>f</code>
 until first use of returned function (reported by Olaf Mersmann).
 </p>
 </li>
 <li> <p><code>quietly=FALSE</code> is now also an optional argument of
 <code>library()</code>, and consequently, <code>quietly</code> is now
 propagated also for loading dependent packages, e.g., in
 <code>require(*, quietly=TRUE)</code>.
 </p>
 </li>
 <li><p> If the loop variable in a <code>for</code> loop was deleted, it
 would be recreated as a global variable.  (Reported by Radford
 Neal; the fix includes his optimizations as well.)
 </p>
 </li>
 <li><p> Task callbacks could report the wrong expression when
 the task involved parsing new code. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14368">PR#14368</a>)
 </p>
 </li>
 <li> <p><code>getNamespaceVersion()</code> failed; this was an
 accidental change in 2.11.0. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14374">PR#14374</a>)
 </p>
 </li>
 <li> <p><code>identical()</code> returned <code>FALSE</code> for external
 pointer objects even when the pointer addresses were the same.
 </p>
 </li>
 <li> <p><code>L$a@x[] &lt;- val</code> did not duplicate in a case it should have.
 </p>
 </li>
 <li> <p><code>tempfile()</code> now always gives a random file name (even
 if the directory is specified) when called directly after
 startup and before the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> RNG had been used.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14381">PR#14381</a>)
 </p>
 </li>
 <li> <p><code>quantile(type=6)</code> behaved inconsistently.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14383">PR#14383</a>)
 </p>
 </li>
 <li> <p><code>backSpline(.)</code> behaved incorrectly when the knot
 sequence was decreasing.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14386">PR#14386</a>)
 </p>
 </li>
 <li><p> The reference BLAS included in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> was assuming that
 <code>0*x</code> and <code>x*0</code> were always zero (whereas they could be
 <code>NA</code> or <code>NaN</code> in IEC 60559 arithmetic).  This was seen
 in results from <code>tcrossprod</code>, and for example that
 <code>log(0) %*% 0</code> gave <code>0</code>.
 </p>
 </li>
 <li><p> The calculation of whether text was completely outside
 the device region (in which case, you draw nothing)
 was wrong for screen devices (which have [0, 0] at top-left).
 The symptom was (long) text disappearing when
 resizing a screen window (to make it smaller).  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14391">PR#14391</a>)
 </p>
 </li>
 <li> <p><code>model.frame(drop.unused.levels = TRUE)</code> did not
 take into account <code>NA</code> values of factors when deciding to
 drop levels. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14393">PR#14393</a>)
 </p>
 </li>
 <li> <p><code>library.dynam.unload</code> required an absolute path for
 <code>libpath</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14385">PR#14385</a>)
 </p>
 <p>Both <code>library()</code> and <code>loadNamespace()</code> now record
 absolute paths for use by <code>searchpaths()</code> and
 <code>getNamespaceInfo(ns, "path")</code>.
 </p>
 </li>
 <li><p> The self-starting model <code>NLSstClosestX</code> failed if some
 deviation was exactly zero.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14384">PR#14384</a>)
 </p>
 </li>
 <li> <p><code>X11(type = "cairo")</code> (and other devices such as
 <code>png</code> using cairographics) and which use Pango font selection
 now work around a bug in Pango when very small fonts (those with
 sizes between 0 and 1 in Pango's internal units) are requested.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14369">PR#14369</a>)
 </p>
 </li>
 <li><p> Added workaround for the font problem with <code>X11(type =
       "cairo")</code> and similar on Mac OS X whereby italic and bold styles
 were interchanged.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=13463">PR#13463</a> amongst many other reports.)
 </p>
 </li>
 <li> <p><code>source(chdir = TRUE)</code> failed to reset the working
 directory if it could not be determined &ndash; that is now an error.
 </p>
 </li>
 <li><p> Fix for crash of <code>example(rasterImage)</code> on
 <code>x11(type="Xlib")</code>.
 </p>
 </li>
 <li><p> Force Quartz to bring the on-screen display up-to-date
 immediately before the snapshot is taken by <code>grid.cap()</code>
 in the Cocoa implementation. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14260">PR#14260</a>)
 </p>
 </li>
 <li> <p><code>model.frame</code> had an unstated 500 byte limit on
 variable names.  (Example reported by Terry Therneau.)
 </p>
 </li>
 <li><p> The 256-byte limit on names is now documented.
 </p>
 </li>
 <li><p> Subassignment by <code>[</code>, <code>[[</code> or <code>$</code> on an
 expression object with value <code>NULL</code> coerced the object to a
 list.
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.11.1 patched</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>install.packages()</code> has a new optional argument
 <code>INSTALL_opts</code> which can be used to pass options to
 <code>R CMD INSTALL</code> for source-package installs.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now runs the package-specific tests
 with <span class="env">LANGUAGE=en</span> to facilitate comparison to
 &lsquo;<span class="file">.Rout.save</span>&rsquo; files.
 </p>
 </li>
 <li> <p><code>sessionInfo()</code> gives more detailed platform
 information, including 32/64-bit and the sub-architecture if one
 is used.
 </p>
 </li></ul>


 <h4>DEPRECATED &amp; DEFUNCT</h4>


 <ul>
 <li><p> The use of Perl configuration variables for <code>R CMD
 	check</code> (as previously documented in &lsquo;Writing R
 Extensions&rsquo;) is deprecated and will be removed in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.12.0.  Use
 the environment variables documented in &lsquo;R Internals&rsquo;
 instead.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>R CMD Rd2dvi</code> failed if run from a path containing
 space(s).  This also affected <code>R CMD check</code>, which calls
 <code>Rd2dvi</code>.
 </p>
 </li>
 <li> <p><code>stripchart()</code> could fail with an empty factor
 level.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14317">PR#14317</a>)
 </p>
 </li>
 <li><p> Text help rendering of <span class="samp">\tabular{}</span> has been
 improved: under some circumstances leading blank columns were
 not rendered.
 </p>
 </li>
 <li> <p><code>strsplit(x, fixed=TRUE)</code> marked UTF-8 strings with the
 local encoding when no splits were found.
 </p>
 </li>
 <li> <p><code>weighted.mean(NA, na.rm=TRUE)</code> and similar now returns
 <code>NaN</code> again, as it did prior to <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.10.0.
 </p>
 </li>
 <li> <p><code>R CMD</code> had a typo in its detection of whether the
 environment variable <span class="env">TEXINPUTS</span> was set (reported by Martin
 Morgan).
 </p>
 </li>
 <li><p> The command-line parser could mistake
 <span class="option">--file=size...</span> for one of the options for setting limits
 for Ncells or Vcells.
 </p>
 </li>
 <li><p> The internal <code>strptime()</code> could corrupt its copy of the
 timezone which would then lead to spurious warnings. (<a href="https://bugs.R-project.org/show_bug.cgi?id=14338">PR#14338</a>)
 </p>
 </li>
 <li> <p><code>dir.create(recursive = TRUE)</code> could fail if one of the
 components existed but was a directory on a read-only file system.
 (Seen on Solaris, where the error code returned is not even listed
 as possible on the man page.)
 </p>
 </li>
 <li><p> The <code>postscript()</code> and <code>pdf()</code> devices will now
 allow <code>lwd</code> values less than 1 (they used to force such
 values to be 1).
 </p>
 </li>
 <li><p> Fixed font face for CID fonts in <code>pdf()</code> graphics output.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14326">PR#14326</a>)
 </p>
 </li>
 <li> <p><code>GERaster()</code> now checks for width or height of zero and
 does nothing in those cases; previously the behaviour was
 undefined, probably device-specific, and possibly dangerous.
 </p>
 </li>
 <li> <p><code>wilcox.test(x, y, conf.int = TRUE)</code> failed with an unhelpful
 message if <code>x</code> and <code>y</code> were constant vectors, and
 similarly in the one-sample case.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14329">PR#14329</a>)
 </p>
 </li>
 <li><p> Improperly calling <code>Recall()</code> from outside a function
 could cause a segfault.  (Reported by Robert McGehee.)
 </p>
 </li>
 <li> <p><span class="samp">\Sexpr[result=rd]</span> in an Rd file added a
 spurious newline, which was displayed as extra whitespace when
 rendered.
 </p>
 </li>
 <li> <p><code>require(save = TRUE)</code> recorded the names of packages
 it failed to load.
 </p>
 </li>
 <li> <p><code>packageStatus()</code> could return a data frame with
 duplicate row names which could then not be printed.
 </p>
 </li>
 <li> <p><code>txtProgressBar(style = 2)</code> did not work correctly.
 </p>
 <p><code>txtProgressBar(style = 3)</code> did not display until a
 non-minimum value was set.
 </p>
 </li>
 <li> <p><code>contour()</code> did not display dashed line types
 properly when contour lines were labelled. (Reported by
 David B. Thompson.)
 </p>
 </li>
 <li> <p><code>tools::undoc()</code> again detects undocumented data
 objects.  Of course, this also affects <code>R CMD check</code>.
 </p>
 </li>
 <li> <p><code>ksmooth(x,NULL)</code> no longer segfaults.
 </p>
 </li>
 <li> <p><code>approxfun()</code>, <code>approx()</code>, <code>splinefun()</code>
 and <code>spline()</code> could be confused by <code>x</code> values that
 were different but so close as to print identically.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14377">PR#14377</a>)
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.11.1</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>R CMD INSTALL</code> checks if dependent packages are
 available early on in the installation of source packages,
 thereby giving clearer error messages.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL --build</code> now names the file in the
 format used for Mac OS X binary files on that platform.
 </p>
 </li>
 <li> <p><code>BIC()</code> in package <span class="pkg">stats4</span> now also works with
 multiple fitted models, analogously to <code>AIC()</code>.
 </p>
 </li></ul>


 <h4>DEPRECATED &amp; DEFUNCT</h4>


 <ul>
 <li><p> Use of file extension &lsquo;<span class="file">.C</span>&rsquo; for C++ code in packages is now
 deprecated: it has caused problems for some <code>make</code>s on
 case-insensitive file systems (although it currently works
 with the recommended toolkits).
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> Command <code>gnutar</code> is preferred to <code>tar</code>
 when configure sets <span class="env">TAR</span>.  This is needed on Mac OS 10.6,
 where the default <code>tar</code>, <code>bsdtar 2.6.2</code>, has been
 reported to produce archives with illegal extensions to tar
 (according to the POSIX standard).
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li><p> The C function <code>mkCharLenCE</code> now no longer reads past
 <code>len</code> bytes (unlikely to be a problem except in user
 code).  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14246">PR#14246</a>)
 </p>
 </li>
 <li><p> On systems without any default <span class="env">LD_LIBRARY_PATH</span> (not
 even &lsquo;<span class="file">/usr/local/lib</span>&rsquo;), <span class="env">[DY]LIB_LIBRARY_PATH</span> is now
 set without a trailing colon.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=13637">PR#13637</a>)
 </p>
 </li>
 <li><p> More efficient implementation of <code>utf8ToInt()</code> on long
 multi-byte strings with many multi-byte characters.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14262">PR#14262</a>)
 </p>
 </li>
 <li> <p><code>aggregate.ts()</code> gave platform-dependent results due to
 rounding error for <code>ndeltat != 1</code>.
 </p>
 </li>
 <li> <p><code>package.skeleton()</code> sometimes failed to fix filenames
 for &lsquo;<span class="file">.R</span>&rsquo; or &lsquo;<span class="file">.Rd</span>&rsquo; files to start with an alphanumeric.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14253">PR#14253</a>)
 </p>
 <p>It also failed when only an S4 class without any methods was
 defined.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14280">PR#14280</a>)
 </p>
 </li>
 <li> <p><code>splinefun(method = "monoH.FC")</code> was not
 quite monotone in rare cases.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14215">PR#14215</a>)
 </p>
 </li>
 <li> <p><code>Rhttpd</code> no longer crashes due to <code>SIGPIPE</code> when
 the client closes the connection prematurely.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14266">PR#14266</a>)
 </p>
 </li>
 <li> <p><code>format.POSIXlt()</code> could cause a stack overflow and
 crash when used on very long vectors.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14267">PR#14267</a>)
 </p>
 </li>
 <li> <p><code>Rd2latex()</code> incorrectly escaped special characters in
 <span class="samp">\usage</span> sections.
 </p>
 </li>
 <li> <p><code>mcnemar.test()</code> could alter the levels (dropping
 unused levels) if passed <code>x</code> and <code>y</code> as factors
 (reported by Greg Snow).
 </p>
 </li>
 <li> <p><code>Rd2pdf</code> sometimes needed a further
 <code>pdflatex</code> pass to get hyperlinked pages correct.
 </p>
 </li>
 <li> <p><code>interaction()</code> produced malformed results when levels were
 duplicated, causing segfaults in <code>split()</code>.
 </p>
 </li>
 <li> <p><code>cut(d, breaks = &lt;n&gt;)</code> now also works for <code>"Date"</code>
 or <code>"POSIXt"</code> argument <code>d</code>.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14288">PR#14288</a>)
 </p>
 </li>
 <li> <p><code>memDecompress()</code> could decompress incompletely rare
 <code>xz</code>-compressed input due to incorrect documentation of
 <code>xz utils</code>.  (Report and patch from Olaf Mersmann.)
 </p>
 </li>
 <li><p> The S4 <code>initialize()</code> methods for <code>"matrix"</code>,
 <code>"array"</code>, and <code>"ts"</code> have been fixed to call
 <code>validObject()</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14284">PR#14284</a>)
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> now behaves the same way with or
 without <span class="option">--no-multiarch</span> on platforms with only one
 installed architecture.  (It used to clean the &lsquo;<span class="file">src</span>&rsquo;
 directory without <span class="option">--no-multiarch</span>.)
 </p>
 </li>
 <li> <p><code>[&lt;-.data.frame</code> was not quite careful enough in
 assigning (and potentially deleting) columns
 right-to-left.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14263">PR#14263</a>)
 </p>
 </li>
 <li> <p><code>rbeta(n, a, b)</code> no longer occasionally returns
 <code>NaN</code> for <code>a &gt;&gt; 1 &gt; b</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14291">PR#14291</a>)
 </p>
 </li>
 <li> <p><code>pnorm(x, log.p = TRUE)</code> could return <code>NaN</code> not
 <code>-Inf</code> for <code>x</code> near (minus for <code>lower.tail=TRUE</code>)
 the largest representable number.
 </p>
 </li>
 <li><p> Compressed data files &lsquo;<span class="file">*.(txt|tab|csv).(gz|bz2|xz)</span>&rsquo;
 were not recognized for the list of data topics and hence for
 packages using <code>LazyData</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14273">PR#14273</a>)
 </p>
 </li>
 <li> <p><code>textConnection()</code> did an unnecessary translation on
 strings in a foreign encoding (e.g., UTF-8 strings on Windows) and
 so was slower than it could have been on very long input strings.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14286">PR#14286</a>)
 </p>
 </li>
 <li> <p><code>tools::Rd2txt()</code> did not render poorly written Rd
 files consistently with other renderers.
 </p>
 <p>It computed widths of strings that would be <code>print()</code>ed with
 escapes incorrectly, for example in the computation of column
 width for <span class="samp">\tabular</span>.
 </p>
 </li>
 <li> <p><code>na.action()</code> did not extract the <code>na.action</code>
 component as documented.
 </p>
 </li>
 <li> <p><code>do.call()</code>ing <code>NextMethod</code> in erroneous ways no
 longer segfaults. (<a href="https://bugs.R-project.org/show_bug.cgi?id=13487">PR#13487</a>)
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.11.0</h3>


 <h4>SIGNIFICANT USER-VISIBLE CHANGES</h4>


 <ul>
 <li><p> Packages must have been installed under <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> &gt;= 2.10.0, as the
 current help system is the only one now supported.
 </p>
 </li>
 <li><p> A port to 64-bit Windows is now available as well as binary
 package repositories: see the &lsquo;R Administration and
 Installation Manual&rsquo;.
 </p>
 </li>
 <li><p> Argument matching for primitive functions is now done in the
 same way as for interpreted functions except for the deliberate
 exceptions
 </p>
 <pre>
 	 call switch .C .Fortran .Call .External
        </pre>
 <p>all of which use positional matching for their first argument,
 and also some internal-use-only primitives.
 </p>
 </li>
 <li><p> The default device for command-line <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> at the console on Mac OS X
 is now <code>quartz()</code> and not <code>X11()</code>.
 </p>
 </li></ul>


 <h4>NEW FEATURES</h4>


 <ul>
 <li><p> The <code>open</code> modes for connections are now interpreted
 more consistently.  <code>open = "r"</code> is now equivalent to
 <code>open = "rt"</code> for all connections.  The default <code>open =
        ""</code> now means <code>"rt"</code> for all connections except the
 compressed-file connections <code>gzfile()</code>, <code>bzfile()</code> and
 <code>xzfile()</code> for which it means <code>"rb"</code>.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> now uses the internal <code>untar()</code>
 function in package <span class="pkg">utils</span>: this ensures that all platforms
 can install <code>bzip2</code>- and <code>xz</code>-compressed
 tarballs.  In case this causes problems (as it has on some
 Windows file systems when run from Cygwin tools) it can be
 overridden by the environment variable <span class="env">R_INSTALL_TAR</span>:
 setting this to a modern external tar program will speed up
 unpacking of large (tens of Mb or more) tarballs.
 </p>
 </li>
 <li> <p><code>help(try.all.packages = TRUE)</code> is much faster
 (although the time taken by the OS to find all the packages the
 first time it is used can dominate the time).
 </p>
 </li>
 <li> <p><code>R CMD check</code> has a new option <span class="option">--timings</span>
 to record per-example timings in file
 &lsquo;<span class="file">&lt;pkg&gt;.Rcheck/&lt;pkg&gt;-Ex.timings</span>&rsquo;.
 </p>
 </li>
 <li><p> The TRE library has been updated to version 0.8.0 (minor bugfixes).
 </p>
 </li>
 <li> <p><code>grep[l]</code>, <code>[g]sub</code> and <code>[g]regexpr</code> now work
 in bytes in an 8-bit locales if there is no marked UTF-8 input
 string: this will be somewhat faster, and for <code>[g]sub()</code>
 give the result in the native encoding rather than in UTF-8
 (which returns to the behaviour prior to <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.10.0).
 </p>
 </li>
 <li><p> A new argument <code>skipCalls</code> has been added to
 <code>browser()</code> so that it can report the original context when
 called by other debugging functions.
 </p>
 </li>
 <li><p> More validity checking of UTF-8 and MBCS strings is done by
 <code>agrep()</code> and the regular-expression matching functions.
 </p>
 </li>
 <li><p> The undocumented restriction on <code>gregexpr()</code> to
 <code>length(text) &gt; 0</code> has been removed.
 </p>
 </li>
 <li><p> Package <span class="pkg">tcltk</span> now sends strings to Tcl in UTF-8: this
 means that strings with a marked UTF-8 encoding are supported in
 non-UTF-8 locales.
 </p>
 </li>
 <li><p> The graphics engine now supports rendering of raster (bitmap)
 images, though not all graphics devices can provide (full)
 support.  Packages providing graphics devices (e.g., <code>Cairo</code>,
 <code>RSvgDevice</code>, <code>cairoDevice</code>) will need to be reinstalled.
 </p>
 <p>There is also support in the graphics engine for capturing
 raster images from graphics devices (again not supported
 on all graphics devices).
 </p>
 </li>
 <li> <p><code>R CMD check</code> now also checks if the package and
 namespace can be unloaded: this provides a check of the
 <code>.Last.lib()</code> and <code>.onUnload()</code> hook functions (unless
 <span class="option">--install=fake</span>).
 </p>
 </li>
 <li> <p><code>prop.table(x)</code> now accepts a one-dimensional table for
 <code>x</code>.
 </p>
 </li>
 <li><p> A new function <code>vapply()</code> has been added, based on a
 suggestion from Bill Dunlap.  It requires that a template for the
 function value be specified, and uses it to determine the output
 type and to check for consistency in the function values.
 </p>
 </li>
 <li><p> The main HTML help page now links to a reformatted copy of
 this &lsquo;<span class="file">NEWS</span>&rsquo; file.  (Suggested by Henrik Bengtsson.)  Package
 index files link to the package &lsquo;<span class="file">DESCRIPTION</span>&rsquo; and
 &lsquo;<span class="file">NEWS</span>&rsquo; files and a list of demos when using dynamic help.
 </p>
 </li>
 <li><p> The <code>[</code> method for class <code>"AsIs"</code> allows the next
 method to change the underlying class.  (Wish of Jens
 Oehlschlägel.)
 </p>
 </li>
 <li> <p><code>write.csv[2]</code> no longer allow argument <code>append</code>
 to be changed: as ever, direct calls to <code>write.table()</code> give
 more flexibility as well as more room for error.
 </p>
 </li>
 <li><p> The index page for HTML help for a package now collapses
 multiple signatures for S4 methods into a single entry.
 </p>
 </li>
 <li><p> The use of <code>.required</code> by <code>require()</code> and
 <code>detach()</code> has been replaced by <code>.Depends</code> which is set
 from the <code>Depends</code> field of a package (even in packages with
 namespaces).  By default <code>detach()</code> prevents such
 dependencies from being detached: this can be overridden by the
 argument <code>force</code>.
 </p>
 </li>
 <li> <p><code>bquote()</code> has been extended to work on function
 definitions.  (Wish of
 <a href="https://bugs.R-project.org/show_bug.cgi?id=14031">PR#14031</a>).
 </p>
 </li>
 <li> <p><code>detach()</code> when applied to an object other than a
 package returns the environment that has been detached, to
 parallel <code>attach()</code>.
 </p>
 </li>
 <li> <p><code>readline()</code> in non-interactive use returns <code>""</code>
 and does not attempt to read from the &lsquo;terminal&rsquo;.
 </p>
 </li>
 <li><p> New function <code>file_ext()</code> in package <span class="pkg">tools</span>.
 </p>
 </li>
 <li> <p><code>xtfrm()</code> is now primitive and internally generic, as
 this allows S4 methods to be set on it without name-space scoping
 issues.
 </p>
 <p>There are now <code>"AsIs"</code> and <code>"difftime"</code> methods, and
 the default method uses <code>unclass(x)</code> if <code>is.numeric(x)</code>
 is true (which will be faster but relies on <code>is.numeric()</code>
 having been set correctly for the class).
 </p>
 </li>
 <li> <p><code>is.numeric(x)</code> is now false for a <code>"difftime"</code>
 object (multiplication and division make no sense for such
 objects).
 </p>
 </li>
 <li><p> The default method of <code>weighted.mean(x, w)</code> coerces
 <code>w</code> to be numeric (aka double); previously only integer
 weights were coerced.  Zero weights are handled specially so an
 infinite value with zero weight does not force an <code>NaN</code>
 result.
 </p>
 <p>There is now a <code>"difftime"</code> method.
 </p>
 </li>
 <li> <p><code>bug.report()</code> now has arguments <code>package</code> and
 <code>lib.loc</code> to generate bug reports about packages.  When
 this is used, it looks for a <code>BugReports</code> field in the
 package &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file, which will be assumed to be a
 URL at which to submit the report, and otherwise generates an
 email to the package maintainer.  (Suggested by Barry
 Rowlingson.)
 </p>
 </li>
 <li> <p><code>quantile()</code> now has a method for the date-time class
 <code>"POSIXt"</code>, and types 1 and 3 (which never interpolate) work
 for Dates and ordered factors.
 </p>
 </li>
 <li> <p><code>length(&lt;POSIXlt&gt;)</code> now returns the length of the
 corresponding abstract timedate-vector rather than always 9 (the
 length of the underlying list structure).  (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14073">PR#14073</a>
 and <a href="https://bugs.R-project.org/show_bug.cgi?id=10507">PR#10507</a>.)
 </p>
 </li>
 <li><p> The <code>readline</code> completion backend no longer sorts
 possible completions alphabetically (e.g., function argument
 names) if <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> was built with <code>readline &gt;= 6</code>.
 </p>
 </li>
 <li> <p><code>select.list()</code> gains a <code>graphics</code> argument to
 allow Windows/Mac users to choose the text interface.  This
 changes the behaviour of <code>new.packages(ask=TRUE)</code> to be like
 <code>update.packages(ask=TRUE)</code> on those platforms in using a
 text menu: use ask=<code>"graphics"</code> for a graphical menu.
 </p>
 </li>
 <li><p> New function <code>chooseBioCmirror()</code> to set the
 <code>"BioC_mirror"</code> option.
 </p>
 </li>
 <li><p> The <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> grammar now prevents using the argument <code>name</code>
 in signatures of S4 methods for <code>$</code> and <code>$&lt;-</code>, since
 they will always be called with a character string value for
 <code>name</code>.  The implicit S4 generic functions have been
 changed to reflect this: packages which included <code>name</code> in
 the signature of their methods need to be updated and
 re-installed.
 </p>
 </li>
 <li><p> The handling of the <code>method</code> argument of <code>glm()</code>
 has been refined following suggestions by Ioannis Kosmidis and
 Heather Turner.
 </p>
 </li>
 <li> <p><code>str()</code> gains a new argument <code>list.len</code> with
 default 99, limiting the number of <code>list()</code> items (per
 level), thanks to suggestions from David Winsenius.
 </p>
 </li>
 <li><p> Having formal arguments of an S4 method in a different order
 from the generic is now an error (the warning having been
 ignored by some package maintainers for a long time).
 </p>
 </li>
 <li><p> New functions <code>enc2native()</code> and <code>enc2utf8()</code>
 convert character vectors with possibly marked encodings to the
 current locale and UTF-8 respectively.
 </p>
 </li>
 <li><p> Unrecognized escapes and embedded nuls in character strings are
 now an error, not just a warning.  Thus option <code>"warnEscapes"</code>
 is no longer needed.  <code>rawToChar()</code> now removes trailing nuls
 silently, but other embedded nuls become errors.
 </p>
 </li>
 <li><p> Informational messages about masked objects displayed when a
 package is attached are now more compact, using <code>strwrap()</code>
 instead of one object per line.
 </p>
 </li>
 <li> <p><code>print.rle()</code> gains argument <code>prefix</code>.
 </p>
 </li>
 <li> <p><code>download.file()</code> gains a <code>"curl"</code> method, mainly
 for use on platforms which have <code>curl</code> but not
 <code>wget</code>, but also for some hard-to-access URLs.
 </p>
 </li>
 <li><p> In <code>Rd</code>, <span class="samp">\eqn</span> and <span class="samp">\deqn</span> will render in
 HTML (and convert to text) upper- and lower-case Greek letters
 (entered as <span class="samp">\alpha</span> ...), <span class="samp">\ldots</span>, <span class="samp">\dots</span>,
 <span class="samp">\ge</span> and <span class="samp">\le</span>.
 </p>
 </li>
 <li> <p><code>utf8ToInt()</code> and <code>intToUtf8()</code> now map <code>NA</code> inputs
 to <code>NA</code> outputs.
 </p>
 </li>
 <li> <p><code>file()</code> has a new argument <code>raw</code> which may help
 if it is used with something other than a regular file, e.g. a
 character device.
 </p>
 </li>
 <li><p> New function <code>strtoi()</code>, a wrapper for the C function
 <code>strtol</code>.
 </p>
 </li>
 <li> <p><code>as.octmode()</code> and <code>as.hexmode()</code> now allow inputs
 of length other than one.
 </p>
 <p>The <code>format()</code> and <code>print()</code> methods for
 <code>"octmode"</code> now preserve names and dimensions (as those for
 <code>"hexmode"</code> did).
 </p>
 <p>The <code>format()</code> methods for classes <code>"octmode"</code> and
 <code>"hexmode"</code> gain a <code>width</code> argument.
 </p>
 </li>
 <li> <p><code>seq.int()</code> returns an integer result in some further cases
 where <code>seq()</code> does, e.g. <code>seq.int(1L, 9L, by = 2L)</code>.
 </p>
 </li>
 <li><p> Added <span class="samp">\subsection{}{}</span> macro to Rd syntax, for
 subsections within sections.
 </p>
 </li>
 <li><p> n-dimensional arrays with dimension names can now be indexed
 by an n-column character matrix.  The indices are matched against
 the dimension names.  <code>NA</code> indices are propagated to the
 result.  Unmatched values and <code>""</code> are not allowed and
 result in an error.
 </p>
 </li>
 <li> <p><code>interaction(drop=TRUE)</code> uses less memory (related to
 <a href="https://bugs.R-project.org/show_bug.cgi?id=14121">PR#14121</a>).
 </p>
 </li>
 <li> <p><code>summary()</code> methods have been added to the
 <code>"srcref"</code> and <code>"srcfile"</code> classes, and various
 encoding issues have been cleaned up.
 </p>
 </li>
 <li><p> If option <code>"checkPackageLicense"</code> is set to <code>TRUE</code> (not
 currently the default), users will be asked to agree to
 non-known-to-be-FOSS package licences at first use.
 </p>
 </li>
 <li><p> Checking <code>setAs(a, b)</code> methods only gives a message
 instead of a warning, when one of <code>a</code> or <code>b</code> is
 unknown.
 </p>
 </li>
 <li><p> New function <code>norm()</code> to compute a matrix norm.
 <code>norm()</code> and also <code>backsolve()</code> and <code>sample()</code>
 have implicit S4 generics.
 </p>
 </li>
 <li><p> Files &lsquo;<span class="file">Renviron.site</span>&rsquo; and &lsquo;<span class="file">Rprofile.site</span>&rsquo; can have
 architecture-specific versions on systems with sub-architectures.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now (by default) also checks Rd files
 for auto-generated content in need of editing, and missing
 argument descriptions.
 </p>
 </li>
 <li> <p><code>aggregate()</code> gains a formula method thanks to a
 contribution by Arni Magnusson.  The data frame method now allows
 summary functions to return arbitrarily many values.
 </p>
 </li>
 <li> <p><code>path.expand()</code> now propagates <code>NA</code> values rather
 than converting them to <code>"NA"</code>.
 </p>
 </li>
 <li> <p><code>file.show()</code> now disallows <code>NA</code> values for file names,
 headers, and pager.
 </p>
 </li>
 <li><p> The &lsquo;fuzz&rsquo; used by <code>seq()</code> and <code>seq.int()</code> has
 been reduced from <code>1e-7</code> to <code>1e-10</code>, which should be
 ample for the double-precision calculations used in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.  It
 ensures that the fuzz never comes into play with sequences of
 integers (wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=14169">PR#14169</a>).
 </p>
 </li>
 <li><p> The default value of <code>RSiteSearch(restrict=)</code> has been
 changed to include vignettes but to exclude R-help.  The R-help
 archives available have been split, with a new option of
 <code>"Rhelp10"</code> for those from 2010.
 </p>
 </li>
 <li><p> New function <code>rasterImage()</code> in the <code>graphics</code>
 package for drawing raster images.
 </p>
 </li>
 <li> <p><code>stats:::extractAIC.coxph()</code> now omits aliased terms
 when computing the degrees of freedom (suggestion of Terry
 Therneau).
 </p>
 </li>
 <li> <p><code>cor()</code> and <code>cov()</code> now test for misuse with
 non-numeric arguments, such as the non-bug report <a href="https://bugs.R-project.org/show_bug.cgi?id=14207">PR#14207</a>.
 </p>
 </li>
 <li> <p><code>pchisq(ncp =, log.p = TRUE)</code> is more accurate for
 probabilities near one.  E.g., <code>pchisq(80, 4, ncp=1,
        log.p=TRUE)</code>.  (Maybe what was meant in <a href="https://bugs.R-project.org/show_bug.cgi?id=14216">PR#14216</a>.)
 </p>
 </li>
 <li> <p><code>maintainer()</code> has been added, to give convenient
 access to the name of the maintainer of a package (contributed by
 David Scott).
 </p>
 </li>
 <li> <p><code>sample()</code> and <code>sample.int()</code> allow zero items to
 be sampled from a zero-length input.  <code>sample.int()</code> gains a
 default value <code>size=n</code> to be more similar to
 <code>sample()</code>.
 </p>
 </li>
 <li> <p><code>switch()</code> returned <code>NULL</code> on error (not
 previously documented on the help page): it now does so
 invisibly, analogously to if-without-else.
 </p>
 <p>It is now primitive: this means that argument <code>EXPR</code> is
 always matched to the first argument and there is no danger of
 partial matching to later named arguments.
 </p>
 </li>
 <li><p> Primitive functions <code>UseMethod()</code>, <code>attr()</code>,
 <code>attr&lt;-()</code>, <code>on.exit()</code>, <code>retracemem()</code> and
 <code>substitute()</code> now use standard argument matching (rather
 than positional matching).  This means that all multi-argument
 primitives which are not internal now use standard argument
 matching except where positional matching is desirable (as for
 <code>switch()</code>, <code>call()</code>, <code>.C()</code> ...).
 </p>
 </li>
 <li><p> All the one-argument primitives now check that any name
 supplied for their first argument is a partial match to the
 argument name as documented on the help page: this also
 applies to replacement functions of two arguments.
 </p>
 </li>
 <li> <p><code>base::which()</code> uses a new <code>.Internal</code> function
 when <code>arr.ind</code> is <code>FALSE</code> resulting in a 10x speedup.
 Thanks to Patrick Aboyoun for implementation suggestions.
 </p>
 </li>
 <li><p> Help conversion to text now uses the first part of
 <span class="samp">\enc{}{}</span> markup if it is representable in the current
 output encoding.  On the other hand, conversion to LaTeX with the
 default <code>outputEncoding = "ASCII"</code> uses the second part.
 </p>
 </li>
 <li><p> A new class <code>"listOfMethods"</code> has been introduced to
 represent the methods in a methods table, to replace the
 deprecated class <code>"MethodsList"</code>.
 </p>
 </li>
 <li> <p><code>any()</code> and <code>all()</code> return early if possible.
 This may speed up operations on long vectors.
 </p>
 </li>
 <li> <p><code>strptime()</code> now accepts <code>"%z"</code> (for the offset
 from UTC in the RFC822 format of <code>+/-hhmm</code>).
 </p>
 </li>
 <li><p> The PCRE library has been updated to version 8.02, a bug-fix
 release which also updates tables to Unicode 5.02.
 </p>
 </li>
 <li><p> Functions which may use a graphical <code>select.list()</code>
 (including <code>menu()</code> and <code>install.packages()</code>) now check
 on a Unix-alike that Tk can be started (and not just
 <code>capabilities("tcltk") &amp;&amp; capabilities("X11")</code>).
 </p>
 </li>
 <li><p> The parser no longer marks strings containing octal or hex
 escapes as being in UTF-8 when entered in a UTF-8 locale.
 </p>
 </li>
 <li><p> On platforms with cairo but not Pango (notably Mac OS X) the
 initial default <code>X11()</code> type is set to <code>"Xlib"</code>: this avoids
 several problems with font selection when done by cairo rather
 than Pango (at least on Mac OS X).
 </p>
 </li>
 <li><p> New function <code>arrayInd()</code> such that <code>which(x,
        arr.ind = TRUE)</code> for an array 'x' is now equivalent to
 <code>arrayInd(which(x), dim(x), dimnames(x))</code>.
 </p>
 </li></ul>


 <h4>DEPRECATED &amp; DEFUNCT</h4>


 <ul>
 <li><p> Bundles of packages are defunct.
 </p>
 </li>
 <li> <p><code>stats::clearNames()</code> is defunct: use <code>unname()</code>.
 </p>
 </li>
 <li><p> Basic regular expressions are defunct, and
 <code>strsplit()</code>, <code>grep()</code>, <code>grepl()</code>, <code>sub()</code>,
 <code>gsub()</code>, <code>regexpr()</code> and <code>gregexpr()</code> no longer
 have an <code>extended</code> argument.
 </p>
 </li>
 <li> <p><code>methods::trySilent()</code> is defunct.
 </p>
 </li>
 <li> <p><code>index.search()</code> (which was deprecated in 2.10.0) is no
 longer exported and has a different argument list.
 </p>
 </li>
 <li><p> Use of multiple arguments to <code>return()</code> is now defunct.
 </p>
 </li>
 <li><p> The use of <code>UseMethod()</code> with more than two arguments
 is now defunct.
 </p>
 </li>
 <li><p> In the <span class="pkg">methods</span> package, the <code>"MethodsList"</code>
 metadata objects which had been superseded by hash tables
 (environments) since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.8.0 are being phased out.  Objects of
 this class are no longer assigned or used as metadata by the
 package.
 </p>
 <p><code>getMethods()</code> is now deprecated, with its internal use replaced
 by <code>findMethods()</code> and other changes.  Creating objects from the
 <code>"MethodsList"</code> class is also deprecated.
 </p>
 </li>
 <li><p> Parsing strings containing both octal/hex and Unicode escapes
 now gives a warning and will become an error in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.12.0.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> UTF-8 is now used for the reference manual and package
 manuals.  This requires LaTeX &lsquo;2005/12/01&rsquo; or later.
 </p>
 </li>
 <li> <p><code>configure</code> looks for a POSIX compliant
 <code>tr</code>, Solaris's <code>/usr/ucb/tr</code> having been found
 to cause <code>Rdiff</code> to malfunction.
 </p>
 </li>
 <li> <p><code>configure</code> is now generated with <code>autoconf
        2.65</code>, which works better on recent systems and on Mac OS X.
 </p>
 </li></ul>


 <h4>PACKAGE INSTALLATION</h4>


 <ul>
 <li><p> Characters in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> source which are not translatable to the
 current locale are now handled more tolerantly:  these will be
 converted to hex codes with a warning.  Such characters are
 only really portable if they appear in comments.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> now tests that the installed package
 can be loaded (and backs out the installation if it cannot): this
 can be suppressed by <span class="option">--no-test-load</span>.  This avoids
 installing/updating a package that cannot be used: common causes
 of failures to load are missing/incompatible external software
 and missing/broken dependent packages.
 </p>
 </li>
 <li><p> Package installation on Windows for a package with a
 &lsquo;<span class="file">src</span>&rsquo; directory now checks if a DLL is created unless there
 is a &lsquo;<span class="file">src/Makefile.win</span>&rsquo; file: this helps catch broken
 installations where the toolchain has not reported problems in
 building the DLL.  (Note: this can be any DLL, not just one named
 &lsquo;<span class="file">&lt;pkg-name&gt;.dll</span>&rsquo;.)
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li><p> Using <code>with()</code>, <code>eval()</code> etc with a list with some
 unnamed elements now works.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14035">PR#14035</a>)
 </p>
 </li>
 <li><p> The <code>"quick"</code> dispatch of S4 methods for primitive
 functions was not happening, forcing a search each time.
 (Dispatch for closures was not affected.)  A side effect is that
 default values for arguments in a method that do not have
 defaults in the generic will now be ignored.
 </p>
 </li>
 <li><p> Trying to dispatch S4 methods for primitives during the search
 for inherited methods slows that search down and potentially
 could cause an infinite recursion.  An internal switch was
 added to turn off all such methods from
 <code>findInheritedMethods()</code>.
 </p>
 </li>
 <li> <p><span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> framework installation (on Mac OS X) would not work
 properly if a rogue <code>Resources</code> directory was present at the
 top level.  Such a non-symlink will now be renamed to
 <code>Resources.old</code> (and anything previously named
 <code>Resources.old</code> removed) as part of the framework
 installation process.
 </p>
 </li>
 <li><p> The checks for conforming S4 method arguments could fail
 when the signature of the generic function omitted some of the
 formal arguments (in addition to <code>...</code>).  Arguments omitted
 from the method definition but conforming (per the documentation)
 should now be ignored (treated as <code>"ANY"</code>) in dispatching.
 </p>
 </li>
 <li><p> The computations for S4 method evaluation when <code>...</code>
 was in the signature could fail, treating <code>...</code> as an
 ordinary symbol.  This has been fixed, for the known cases.
 </p>
 </li>
 <li><p> Various <code>ar()</code> fitting methods have more protection for
 singular fits.
 </p>
 </li>
 <li> <p><code>callNextMethod</code> now works again with the <code>drop=</code>
 argument in <span class="samp">[</span>
 </p>
 </li>
 <li> <p><code>parse()</code> and <code>parse_Rd()</code> miscounted columns when
 multibyte UTF-8 characters were present.
 </p>
 </li>
 <li><p> Formatting of help pages has had minor improvements: extra
 blank lines have been removed from the text format, and empty
 package labels removed from HTML.
 </p>
 </li>
 <li> <p><code>cor(A, B)</code> where <code>A</code> is <i>n x 1</i>
 and <code>B</code> a 1-dimensional array segfaulted or gave an internal
 error.  (The case <code>cor(B, A)</code> was
 <a href="https://bugs.R-project.org/show_bug.cgi?id=7116">PR#7116</a>.)
 </p>
 </li>
 <li> <p><code>cut.POSIXt()</code> applied to a start value after the DST
 transition on a DST-change day could give the wrong time for
 argument <code>breaks</code> in units of days or longer.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14208">PR#14208</a>)
 </p>
 </li>
 <li> <p><code>do_par()</code> <code>UNPROTECT</code>ed too early  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14214">PR#14214</a>)
 </p>
 </li>
 <li><p> Subassignment <code>x[[....]] &lt;- y</code> didn't check for a
 zero-length right hand side, and inserted a rubbish
 value.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14217">PR#14217</a>)
 </p>
 </li>
 <li> <p><code>fisher.test()</code> no longer gives a P-value <strong>very</strong>
 slightly &gt; 1, in some borderline cases.
 </p>
 </li>
 <li><p> Internal function <code>matchArgs()</code> no longer modifies the
 general purpose bits of the <code>SEXP</code>s that make up the formals
 list of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> functions.  This fixes an invalid error message that
 would occur when a garbage collection triggered a second call to
 <code>matchArgs</code> for the same function <em>via</em> a finalizer.
 </p>
 </li>
 <li> <p><code>gsub()</code> in 2.10.x could fail from stack overflow for
 extremely long strings due to temporary data being allocated on
 the stack.  Also, <code>gsub()</code> with <code>fixed=TRUE</code> is in some
 circumstances considerably faster.
 </p>
 </li>
 <li><p> Several primitives, including <code>attributes()</code>,
 <code>attr&lt;-()</code> <code>interactive()</code>, <code>nargs()</code> and
 <code>proc.time()</code>, did not check that they were called with the
 correct number of arguments.
 </p>
 </li>
 <li><p> A potential race condition in <code>list.files()</code> when other
 processes are operating on the directory has been fixed; the code
 now dynamically allocates memory for file listings in a single
 pass instead of making an initial count pass.
 </p>
 </li>
 <li> <p><code>mean(x, trim=, na.rm = FALSE)</code> failed to return <code>NA</code> if
 <code>x</code> contained missing values.  (Reported by Bill Dunlap.)
 </p>
 </li>
 <li><p> Extreme tail behavior of, <code>pbeta()</code> {and hence
 <code>pf()</code>}, e.g., <code>pbeta(x, 3, 2200, lower.tail=FALSE,
        log.p=TRUE)</code> now returns finite values instead of jumping to
 <code>-Inf</code> too early.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14230">PR#14230</a>).
 </p>
 </li>
 <li> <p><code>parse(text=x)</code> misbehaved for objects <code>x</code> that
 were not coerced internally to character, notably symbols.
 (Reported to R-devel by Bill Dunlap.)
 </p>
 </li>
 <li><p> The internal C function <code>coerceSymbol</code> now handles
 coercion to character, and warns if coercion fails (rather than
 silently returning <code>NULL</code>).  This allows a name to be given
 where a character vector is required in functions which coerce
 internally.
 </p>
 </li>
 <li><p> The interpretation by <code>strptime()</code> of <code>"%c"</code> was
 non-standard (not that it is ever advisable to use locale- and
 system-specific input formats).
 </p>
 </li>
 <li><p> capabilities(<code>"X11"</code>) now works the same way on Mac OS X as on
 other platforms (and as documented: it was always true for R
 built with <span class="option">--with-aqua</span>, as the CRAN builds are).
 </p>
 </li>
 <li><p> The <code>X11()</code> device with cairo but not Pango (notably Mac OS X)
 now checks validity of text strings in UTF-8 locales (since
 Pango does but cairo it seems does not).
 </p>
 </li>
 <li> <p><code>read.fwf()</code> misread multi-line records when n was specified.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14241">PR#14241</a>)
 </p>
 </li>
 <li> <p><code>all.equal(tolerance = e)</code> passes the numeric
 tolerance also to the comparison of the attributes.
 </p>
 </li>
 <li> <p><code>pgamma(0, 0)</code>, a boundary case, now returns 0, its
 limit from the left, rather than the limit from the right.
 </p>
 </li>
 <li><p> Issuing <code>POST</code> requests to the internal web server could stall
 the request under certain circumstances.
 </p>
 </li>
 <li> <p><code>gzcon( &lt;textConnection&gt; )</code>, an error, no longer
 damages the connection (in a way to have it
 segfault).  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14237">PR#14237</a>)
 </p>
 </li>
 <li><p> All the results from <code>hist()</code> now use the nominal
 <code>breaks</code> not those adjusted by the numeric &lsquo;fuzz&rsquo;:
 in recent versions the nominal <code>breaks</code> were reported but
 the 'density' referred to the intervals used in the calculation
 &ndash; which mattered very slightly for one of the extreme bins.
 (Based on a report by Martin Becker.)
 </p>
 </li>
 <li><p> If <code>xy[z].coords</code> (used internally by many graphics
 functions) are given a list as <code>x</code>, they now check that
 the list has suitable names and give a more informative error
 message.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=13936">PR#13936</a>)
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.10.1 patched</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li><p> The handling of line textures in the <code>postscript()</code> and
 <code>pdf()</code> devices was set up for round end caps (the only type
 which existed at the time): it has now been adjusted for butt
 endcaps.
 </p>
 </li>
 <li> <p><code>lchoose(a, k)</code> is now defined as
 <code>log(abs(choose(a,k)))</code>, analogously to <code>lfactorial()</code>.
 </p>
 </li>
 <li><p> Although <code>\eqn{}</code> in Rd files is defined as a
 &lsquo;verbatim&rsquo; macro, many packages expected <code>\dots</code> and
 <code>\ldots</code> to be interpreted there (as was the case in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> &lt;
 2.10.0), so this is now done (using an ellipsis in HTML
 rendering).
 </p>
 </li>
 <li><p> Escaping of braces in quoted strings in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>-code sections of
 Rd files is allowed again.  This had been changed for the new Rd
 format in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.10.0 but was only documented on the developer site
 and was handled inconsistently by the converters: text and
 example conversion removed the escapes but HTML conversion did
 not.
 </p>
 </li>
 <li><p> The PCRE library has been updated to version 8.01, a bug-fix
 release.
 </p>
 </li>
 <li> <p><code>tools::readNEWS()</code> now accepts a digit as the first
 character of a news section.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li><p> Using <code>read.table(header=TRUE)</code> on a header with an
 embedded new line would copy part of the header into the data.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14103">PR#14103</a>)
 </p>
 </li>
 <li> <p><code>qpois(p = 1, lambda = 0)</code> now gives 0 as for all other
 <code>p</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14135">PR#14135</a>)
 </p>
 </li>
 <li><p> Functions related to string comparison
 (e.g., <code>unique()</code>, <code>match()</code>) could cause crashes when
 used with strings not in the native encoding, e.g., UTF-8 strings
 on Windows.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14114">PR#14114</a> and <a href="https://bugs.R-project.org/show_bug.cgi?id=14125">PR#14125</a>)
 </p>
 </li>
 <li> <p><code>x[ , drop=TRUE]</code> dropped an <code>NA</code> level even if it
 was in use.
 </p>
 </li>
 <li><p> The dynamic HTML help system reported the wrong MIME type
 for the style sheet.
 </p>
 </li>
 <li> <p><code>tools::codoc()</code> (used by <code>R CMD check</code>) was
 missing cases where the function had no arguments but was
 documented to have some.
 </p>
 </li>
 <li><p> Help links containing special characters (e.g., <code>"?"</code>)
 were not generated correctly when rendered in
 HTML.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14155">PR#14155</a>)
 </p>
 </li>
 <li> <p><code>lchoose(a, k)</code> no longer wrongly gives <code>NaN</code> for
 negative <code>a</code>.
 </p>
 </li>
 <li> <p><code>ks.test()</code> could give a p-value that was off by one
 observation due to rounding error.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14145">PR#14145</a>)
 </p>
 </li>
 <li> <p><code>readBin()</code>/<code>readChar()</code> when reading millions of
 character strings in a single call used excessive amounts of
 memory (which also slowed them down).
 </p>
 </li>
 <li> <p><code>R CMD SHLIB</code> could fail if used with paths that
 were not alphanumeric, e.g. contained <code>+</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14168">PR#14168</a>)
 </p>
 </li>
 <li> <p><code>sprintf()</code> was not re-entrant, which potentially
 caused problems if an <code>as.character()</code> method called it.
 </p>
 </li>
 <li><p> The <code>quartz()</code> device did not restore the clipping
 region when filling the background for a new page.  This could be
 observed in multi-page bitmap output as stale outer regions of
 the plot.
 </p>
 </li>
 <li> <p><code>p.adjust(method, n)</code> now works correctly for the
 rare case <code>n &gt; length(p)</code>, also when method differs from
 <code>"bonferroni"</code> or <code>"none"</code>, thanks to a patch from
 Gordon Smyth.
 </p>
 </li>
 <li> <p><code>tools::showNonASCII()</code> failed to detect non-ASCII characters
 if <code>iconv()</code> (incorrectly) converted them to different ASCII
 characters.  (Seen on Windows only.)
 </p>
 </li>
 <li> <p><code>tcrossprod()</code> wrongly failed in some cases when one of the
 arguments was a vector and the other a matrix.
 </p>
 </li>
 <li> <p><code>[cr]bind(..., deparse.level=2)</code> was not always giving
 names when documented to do so.  (Discovered whilst investigating
 <a href="https://bugs.R-project.org/show_bug.cgi?id=14189">PR#14189</a>.)
 </p>
 </li>
 <li> <p><code>match(incomparables=&lt;non-NULL&gt;)</code> could in rare cases
 infinite-loop.
 </p>
 </li>
 <li> <p><code>poisson.test()</code> needed to pass argument
 <code>conf.level</code> to <code>binom.test()</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14195">PR#14195</a>)
 </p>
 </li>
 <li><p> The <code>"nls"</code> method for <code>df.residual()</code> gave
 incorrect results for models fitted with <code>na.action =
        na.exclude</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14194">PR#14194</a>)
 </p>
 </li>
 <li><p> A change to <code>options(scipen=)</code> was only implemented
 when printing next occurred, even though it should have affected
 intervening calls to <code>axis()</code>, <code>contour()</code> and
 <code>filledcontour()</code>.
 </p>
 </li>
 <li> <p><code>prettyNum(drop0trailing=TRUE)</code> did not handle signs of
 imaginary parts of complex numbers correctly (and this was used
 by <code>str()</code>: <a href="https://bugs.R-project.org/show_bug.cgi?id=14201">PR#14201</a>).
 </p>
 </li>
 <li> <p><code>system.time()</code> had the <code>sys.child</code> component
 wrong (copied <code>user.child</code> instead) on systems with
 <code>HAVE_GETRUSAGE</code>.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14210">PR#14210</a>)
 </p>
 </li>
 <li><p> Changing both line texture and line cap (end) resulted in the
 latter to be omitted form the PDF code.  In addition, line cap
 (end) and join are now set explicitly in PDF output to ensure
 correct defaults.
 </p>
 </li>
 <li><p> The suppression of auto-rotation in <code>bitmap()</code> and
 <code>dev2bitmap()</code> with the <code>"pdfwrite"</code> device was not
 working correctly.
 </p>
 </li>
 <li> <p><code>plot(ecdf(), log="x")</code> no longer gives an incorrect
 warning.
 </p>
 </li>
 <li> <p><code>read.fwf()</code> works again when argument <code>file</code> is
 a connection.
 </p>
 </li>
 <li><p> Startup files will now be found if their paths exceed 255
 bytes.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14228">PR#14228</a>)
 </p>
 </li>
 <li> <p><code>contrasts&lt;-</code> (in the <span class="pkg">stats</span> package) no longer has an
 undeclared dependence on <span class="pkg">methods</span> (introduced in 2.10.0).
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.10.1</h3>


 <h4>NEW FEATURES</h4>


 <ul>
 <li><p> The PCRE library has been updated to version 8.00.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> has new options <span class="option">--no-R</span>,
 <span class="option">--no-libs</span>, <span class="option">--no-data</span>, <span class="option">--no-help</span>,
 <span class="option">--no-demo</span>, <span class="option">--no-exec</span>, and <span class="option">--no-inst</span> to
 suppress installation of the specified part of the package.
 These are intended for special purposes (e.g., building a database
 of help pages without fully installing all packages).
 </p>
 </li>
 <li><p> The documented line-length limit of 4095 bytes when reading
 from the console now also applies also to <code>parse(file="")</code>
 (which previously had a limit of around 1024 bytes).
 </p>
 </li>
 <li><p> A Bioconductor mirror can be set for use by
 <code>setRepositories()</code> <em>via</em> the option <code>"BioC_mirror"</code>,
 e.g. the European mirror can be selected by
 <code>options(BioC_mirror="http://bioconductor.statistik.tu-dortmund.de")</code>.
 </p>
 </li>
 <li><p> Double-clicking in a <code>tk_select.list()</code> list box now
 selects the item and closes the list box (as happens on the
 Windows <code>select.list()</code> widget).
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li> <p><code>configure</code> will be able to find a usable
 <code>libtiff</code> in some rare circumstances where it did not
 previously (where <code>libtiff</code> needed to be linked explicitly against
 <code>-ljpeg</code>).
 </p>
 </li>
 <li><p> Making <code>refman.pdf</code> works around a problem with the indexing
 with <code>hyperref 6.79d</code> and later.
 </p>
 </li></ul>


 <h4>DEPRECATED &amp; DEFUNCT</h4>


 <ul>
 <li><p> The <code>extended</code> argument is deprecated in
 <code>strsplit()</code>, <code>grep()</code>, <code>grepl()</code>, <code>sub()</code>,
 <code>gsub()</code>, <code>regexpr()</code> and <code>gregexpr()</code> (not just
 the value <code>extended = FALSE</code>) and will be removed in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 2.11.0.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li> <p><code>trigamma(x)</code> and other <code>psigamma(x, n)</code> calls
 are now accurate for very large <code>abs(x)</code>.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14020">PR#14020</a>)
 </p>
 </li>
 <li> <p><code>[g]sub(perl=FALSE, fixed=FALSE)</code> could use excessive
 stack space when used with a very long vector containing some
 non-ASCII strings.
 </p>
 </li>
 <li><p> The default method of <code>weighted.mean(na.rm = TRUE)</code>
 did not omit weights for <code>NA</code> observations in 2.10.0.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14032">PR#14032</a>)
 </p>
 </li>
 <li> <p><code>[g]regexpr(pattern, fixed = TRUE)</code> returned match
 positions in bytes (not characters) in an MBCS locale if
 <code>pattern</code> was a single byte.
 </p>
 <p><code>[g]sub(fixed = TRUE)</code> with a single-byte pattern could
 conceivably have matched part of a multibyte character in a
 non-UTF-8 MBCS.
 </p>
 </li>
 <li> <p><code>findLineNum()</code> and <code>setBreakpoint()</code> would
 sometimes fail if the specified file was not in the current
 directory.
 </p>
 </li>
 <li><p> Package <span class="pkg">tcltk</span>'s <code>demo(tkdensity)</code> was broken in
 2.9.0 when <code>demo()</code> was changed to set <code>par(ask = TRUE)</code>.
 </p>
 </li>
 <li> <p><code>gsub()</code> with backrefs could fail on extremely long strings
 (hundreds of thousands of characters) due to integer overflow
 in a length calculation.
 </p>
 </li>
 <li> <p><code>abline(untf=TRUE)</code> now uses a better x-grid in
 log-scale, e.g., for <code>plot(c(1,300), c(1,300), log="xy");
 	abline(4,1, untf=TRUE)</code>.
 </p>
 </li>
 <li> <p><code>detach()</code>/<code>unloadNamespace()</code> arrange to flush
 the package's lazyload cache of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> objects once the
 package/namespace is no longer needed.
 </p>
 </li>
 <li><p> There have been small fixes to the rendering of help,
 e.g. <code>\command</code> is now rendered verbatim
 (so e.g. <code>--</code> is not interpreted, <a href="https://bugs.R-project.org/show_bug.cgi?id=14045">PR#14045</a>).
 </p>
 <p>Also, there are many small changes to help files where the new
 converters were not rendering them in the same way as before.
 </p>
 </li>
 <li> <p><code>available.packages()</code> would fail when run on a
 repository with no packages meeting the filtering conditions.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14042">PR#14042</a>)
 </p>
 </li>
 <li> <p><code>rep(x, times, each = 2)</code> gave invalid results when the
 <code>times</code> argument was a vector longer than <code>x</code>.
 Reported by Bill Dunlap.
 </p>
 </li>
 <li><p> An error when <code>unloadNamespace()</code> attempted to run the
 <code>.onUnload()</code> function gave an error in the reporting
 function and so was not reported properly.
 </p>
 </li>
 <li><p> Text help rendering did not handle very long input lines
 properly.
 </p>
 </li>
 <li> <p><code>promptMethods()</code> generated signature documentation
 improperly.
 </p>
 </li>
 <li> <p><code>pgamma(x, a, lower.tail=FALSE)</code> and
 <code>qgamma(...)</code> are now considerably more accurate in some
 regions for very small <code>a</code>.  <code>qgamma()</code> now correctly
 returns <code>0</code> instead of <code>NaN</code> in similar extreme cases,
 and <code>qgamma()</code> no longer warns in the case of small
 <code>a</code>, see (<a href="https://bugs.R-project.org/show_bug.cgi?id=12324">PR#12324</a>).
 </p>
 </li>
 <li> <p><code>unname()</code> now also removes names from a zero length vector.
 </p>
 </li>
 <li><p> Printing results from <code>ls.str()</code> no longer evaluates
 unevaluated calls.
 </p>
 </li>
 <li> <p><code>complete.cases()</code> failed on a 0-column data frame
 argument.  (Underlies <a href="https://bugs.R-project.org/show_bug.cgi?id=14066">PR#14066</a>.)
 </p>
 <p>It could return nonsensical results if no input determined the
 number of cases (seen in the no-segfault tests).
 </p>
 </li>
 <li><p> An error in <code>nls()</code> with a long formula could cause a
 segfault.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14059">PR#14059</a>)
 </p>
 </li>
 <li> <p><code>qchisq(p, df, ncp, lower.tail = FALSE)</code> with
 <code>ncp &gt;= 80</code> was inaccurate for small <code>p</code> (as the help page
 said): it is now less inaccurate.  (In part,
 <a href="https://bugs.R-project.org/show_bug.cgi?id=13999">PR#13999</a>.)
 </p>
 <p>For <code>ncp</code> less than but close to 80, <code>pchisq()</code> and
 <code>qchisq()</code> are more accurate for probabilities very close
 to 1 (a series expansion was truncated slightly too early).
 </p>
 <p><code>pchisq(x, df, ncp)</code> can no longer return values just larger
 than one for large values of <code>ncp</code>.
 </p>
 </li>
 <li> <p><code>intToUtf8()</code> could fail when asked to produce 10Mb or more
 strings, something it was never intended to do: unfortunately
 Windows crashed <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> (other OSes reported a lack of resources).
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=14068">PR#14068</a>)
 </p>
 </li>
 <li> <p><code>chisq.test()</code> could fail when given argument
 <code>x</code> or <code>y</code> which deparsed to more than one line.
 (Reported by Laurent Gauthier.)
 </p>
 </li>
 <li><p> S4 methods are uncached whenever the namespace containing
 them is unloaded (by <code>unloadNamespace()</code> as well as by
 <code>detach(unload = TRUE))</code>.
 </p>
 </li>
 <li><p> The internal record-keeping by
 <code>dyn.load</code>/<code>dyn.unload</code> was incomplete, which could
 crash <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> if a DLL that registered <code>.External</code> routines had
 earlier been unloaded.
 </p>
 </li>
 <li> <p><code>bessel[JY](x, nu)</code> with <code>nu</code> a negative integer (a
 singular case) is now correct, analogously to <code>besselI()</code>,
 see <a href="https://bugs.R-project.org/show_bug.cgi?id=13556">PR#13556</a>.
 </p>
 </li>
 <li> <p><code>tools::file_path_as_absolute()</code> doubled the file
 separator when applied to a file such as &lsquo;<span class="file">&quot;/vmunix&quot;</span>&rsquo; or (on
 Windows) &lsquo;<span class="file">&quot;d:/afile&quot;</span>&rsquo; in a directory for which
 <code>getwd()</code> would return a path with a trailing separator
 (largely cosmetic, as reasonable file systems handle such a path
 correctly).  (Perhaps what was meant by <a href="https://bugs.R-project.org/show_bug.cgi?id=14078">PR#14078</a>.)
 </p>
 </li>
 <li> <p><code>unsplit(drop = TRUE)</code> applied to a data frame failed
 to pass <code>drop</code> to the computation of row names.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14084">PR#14084</a>)
 </p>
 </li>
 <li><p> The <code>"difftime"</code> method of <code>mean()</code> ignored its
 <code>na.rm</code> argument.
 </p>
 </li>
 <li> <p><code>tcltk::tk_select.list()</code> is now more likely to remove the
 widget immediately after selection is complete.
 </p>
 </li>
 <li><p> Adding/subtracting a <code>"difftime"</code> object to/from a
 <code>"POSIXt"</code> or <code>"Date"</code> object works again (it was
 broken by the addition of <code>Ops.difftime</code>).
 </p>
 </li>
 <li><p> Conversion to latex of an Rd file with no aliases failed.
 </p>
 </li>
 <li> <p><code>wilcox.test(conf.int=TRUE)</code> has achieved.level
 corrected and, for <code>exact=FALSE</code>, now returns a
 <code>estimate</code> component which does not depend on the
 alternative used.
 </p>
 </li>
 <li> <p><code>help.search()</code> failed when the package argument was
 specified.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=14113">PR#14113</a>)
 </p>
 </li>
 <li> <p><code>switch(EXPR = "A")</code> now returns <code>NULL</code>, as does
 <code>switch(1)</code> (which used to signal an error).
 </p>
 </li></ul>


 <h3><img class="toplogo" src="../help/figures/../../html/Rlogo.svg" alt="[R logo]" /> CHANGES IN R VERSION 2.10.0</h3>


 <h4>SIGNIFICANT USER-VISIBLE CHANGES</h4>


 <ul>
 <li><p> Package help is now converted from <code>Rd</code> by the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>-based
 converters that were first introduced in 2.9.0.  This means
 </p>

 <ul>
 <li><p> Packages that were installed by R-devel after 2009-08-09
 should not be used with earlier versions of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>, and most
 aspects of package help (including the runnable examples)
 will be missing if they are so used.
 </p>
 </li>
 <li><p> Text, HTML and latex help and examples for packages
 installed under the new system are converted on-demand from
 stored parsed Rd files.  (Conversions stored in packages
 installed under <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> &lt; 2.10.0 are used if no parsed Rd files
 are found.  It is recommended that such packages be
 re-installed.)
 </p>
 </li></ul>

 </li>
 <li><p> HTML help is now generated dynamically using an HTTP server
 running in the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> process and listening on the loopback
 interface.
 </p>

 <ul>
 <li><p> Those worried about security implications of such
 a server can disable it by setting the environment variable
 <span class="env">R_DISABLE_HTTPD</span> to a non-empty value.  This disables
 <code>help.start()</code> and HTML help (so text help is shown
 instead).
 </p>
 </li>
 <li><p> The Java/Javascript search engine has been replaced by an
 HTML interface to <code>help.search()</code>.  <code>help.start()</code> no
 longer has an argument <code>searchEngine</code> as it is no longer
 needed.
 </p>
 </li>
 <li><p> The HTML help can now locate cross-references of the form
 <code>\link[pkg]{foo}</code> and <code>\link[pkg:foo]{bar}</code>
 where <code>foo</code> is an alias in the package, rather than the
 documented (basename of a) filename (since the documentation
 has been much ignored).
 </p>
 </li></ul>

 </li></ul>


 <h4>NEW FEATURES</h4>


 <ul>
 <li> <p><code>polygon()</code>, <code>pdf()</code> and <code>postscript()</code> now
 have an argument <code>fillOddEven</code> (default <code>FALSE</code>),
 which controls the mode used for polygon fills of
 self-intersecting shapes.
 </p>
 </li>
 <li><p> New <code>debugonce()</code> function; further,
 <code>getOption("deparse.max.lines")</code> is now observed when
 debugging, from a code suggestion by John Brzustowski.
 (<a href="https://bugs.R-project.org/show_bug.cgi?id=13647">PR#13647</a>/8)
 </p>
 </li>
 <li> <p><code>plot()</code> methods for <code>"stepfun"</code> and hence
 <code>"ecdf"</code> no longer plot points by default for <code>n &gt;= 1000</code>.
 </p>
 </li>
 <li> <p><code>[g]sub(perl=TRUE)</code> now also supports <code>"\E"</code>
 in order to <strong>end</strong> <code>"\U"</code> and <code>"\L"</code> case changes,
 thanks to a patch from Bill Dunlap.
 </p>
 </li>
 <li> <p><code>factor()</code>, <code>levels()&lt;-</code>, etc, now ensure that
 the resulting factor levels are unique (as was always the implied
 intention).  Factors with duplicated levels are still
 constructible by low-level means, but are now declared illegal.
 </p>
 </li>
 <li><p> New <code>print()</code> (S3) method for class <code>"function"</code>,
 also used for auto-printing.  Further, <code>.Primitive</code>
 functions now print and auto-print identically.  The new method
 is based on code suggestions by Romain François.
 </p>
 </li>
 <li><p> The <code>print()</code> and <code>toLatex()</code> methods for class
 <code>"sessionInfo"</code> now show the locale in a nicer format and
 have arguments to suppress locale information.
 </p>
 </li>
 <li><p> In addition to previously only <code>round()</code>, there are
 other <code>Math</code> group (S3) methods for <code>"difftime"</code>,
 such as <code>floor()</code>, <code>signif()</code>, <code>abs()</code>, etc.
 </p>
 </li>
 <li><p> For completeness, <code>old.packages()</code> and
 <code>available.packages()</code> allow arguments <code>type</code> to be
 specified (you could always specify arguments <code>available</code>
 or <code>contriburl</code>).
 </p>
 </li>
 <li> <p><code>available.packages()</code> by default only returns
 information on the latest versions of packages whose version
 requirements are satisfied by the currently running <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>.
 </p>
 </li>
 <li> <p><code>tools::write_PACKAGES()</code> has a new argument
 <code>latestOnly</code>, which defaults to <code>TRUE</code> when only the
 latest versions in the repository will be listed in the index.
 </p>
 </li>
 <li> <p><code>getOption()</code> has a new argument <code>default</code> that
 is returned if the specified option is not set.  This simplifies
 querying a value and checking whether it is <code>NULL</code> or not.
 </p>
 </li>
 <li> <p><code>parse()</code> now warns if the requested encoding is not supported.
 </p>
 </li>
 <li><p> The <code>"table"</code> method of <code>as.data.frame()</code> gains a
 <code>stringsAsFactors</code> argument to allow the classifying
 factors to be returned as character vectors rather than the
 default factor type.
 </p>
 </li>
 <li><p> If <code>model.frame.default()</code> encounters a character
 variable where <code>xlev</code> indicates a factor, it now converts
 the variable to a factor (with a warning).
 </p>
 </li>
 <li> <p><code>curve()</code> now returns a list containing the points that
 w§ere drawn.
 </p>
 </li>
 <li> <p><code>spineplot()</code> now accepts <code>axes = FALSE</code>, for
 consistency with other functions called by <code>plot.factor()</code>.
 </p>
 </li>
 <li><p> The Kendall and Spearman methods of <code>cor.test()</code> can
 optionally use continuity correction when not computing exact
 p-values.  (The Kendall case is the wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=13691">PR#13691</a>.)
 </p>
 </li>
 <li> <p><span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> now keeps track of line numbers during execution for code
 sourced with <code>options(keep.source = TRUE)</code>.  The source
 reference is displayed by debugging functions such as
 <code>traceback()</code>, <code>browser()</code>, <code>recover()</code>, and
 <code>dump.frames()</code>, and is stored as an attribute on each
 element returned by <code>sys.calls()</code>.
 </p>
 </li>
 <li><p> More functions now have an implicit (S4) generic definition.
 </p>
 </li>
 <li> <p><code>quantile.default()</code> now disallows factors (wish of
 <a href="https://bugs.R-project.org/show_bug.cgi?id=13631">PR#13631</a>) and its help
 documents what numeric-like properties its input need to have to
 work correctly.
 </p>
 </li>
 <li> <p><code>weighted.mean()</code> is now generic and has <code>"Date"</code>,
 <code>"POSIXct"</code> and <code>"POSIXlt"</code> methods.
 </p>
 </li>
 <li><p> Naming subscripts (e.g., <code>x[i=1, j=2]</code>) in
 <code>data.frame</code> methods for <code>[</code> and <code>[[</code> now gives a
 warning.  (Names are ignored in the default method, but could
 have odd semantics for other methods, and do for the data.frame
 ones.)
 </p>
 </li>
 <li> <p><code>as.data.frame()</code> has an <code>"aovproj"</code> method.
 (Wish of <a href="https://bugs.R-project.org/show_bug.cgi?id=13505">PR#13505</a>)
 </p>
 </li>
 <li> <p><code>as.character(x)</code> for numeric <code>x</code> no longer
 produces strings such as <code>"0.30"</code>, i.e., with trailing
 zeros.  This change also renders levels construction in
 <code>factor()</code> more consistent.
 </p>
 </li>
 <li> <p><code>codocClasses()</code>, which checks consistency of the
 documentation of S4 class slots, now does so in considerably more
 cases.  The documentation of inherited slots (from superclasses)
 is now optional.  This affects <code>R CMD check &lt;pkg&gt;</code> when
 the package defines S4 classes.
 </p>
 </li>
 <li> <p><code>codoc()</code> now also checks S4 methods for code/documentation
 mismatches.
 </p>
 </li>
 <li> <p><code>for()</code>, <code>while()</code>, and <code>repeat()</code> loops now
 always return NULL as their (invisible) value.  This change was
 needed to address a reference counting bug without creating
 performance penalties for some common use cases.
 </p>
 </li>
 <li><p> The <code>print()</code> method for <code>ls.str()</code> results now
 obeys an optional <code>digits</code> argument.
 </p>
 </li>
 <li><p> The <code>method</code> argument of <code>glm()</code> now allows
 user-contributed methods.
 </p>
 </li>
 <li><p> More general <code>reorder.default()</code> replaces functionality of
 <code>reorder.factor()</code> and <code>reorder.character()</code>.
 </p>
 </li>
 <li><p> The function <code>aspell()</code> has been added to provide an
 interface to the <code>Aspell</code> spell-checker.
 </p>
 </li>
 <li><p> Filters <code>RdTextFilter()</code> and <code>SweaveTeXFilter()</code>
 have been added to the tools package to provide support for
 <code>aspell()</code> or other spell checkers.
 </p>
 </li>
 <li> <p><code>xtabs()</code> with the new argument <code>sparse = TRUE</code>
 now returns a sparse Matrix, using package <a href="https://CRAN.R-project.org/package=Matrix"><span class="pkg">Matrix</span></a>.
 </p>
 </li>
 <li> <p><code>contr.sum()</code> etc gain an argument <code>sparse</code>
 which allows sparse matrices to be returned.
 </p>
 <p><code>contrasts()</code> also gains a <code>sparse</code> argument which it
 passes to the actual contrast function if that has a formal
 argument <code>sparse</code>.
 </p>
 <p><code>contrasts(f, .) &lt;- val</code> now also works when <code>val</code>
 is a sparse Matrix.  It is planned that <code>model.matrix()</code>
 will work with such factors <code>f</code> in the future.
 </p>
 </li>
 <li> <p><code>readNEWS()</code> will recognize a UTF-8 byte-order mark
 (BOM) in the &lsquo;<span class="file">NEWS</span>&rsquo; file.  However, it is safer to use only
 ASCII code there because not all editors recognize BOMs.
 </p>
 </li>
 <li><p> New utility function <code>inheritedSlotNames()</code> for S4
 class programming.
 </p>
 </li>
 <li> <p><code>tabulate()</code> now allows NAs to pass through (and be ignored).
 </p>
 </li>
 <li><p> If <code>debug()</code> is called on an S3 generic function then
 all methods are debugged as well.
 </p>
 </li>
 <li><p> Outlier symbols drawn by <code>boxplot()</code> now obey the
 <code>outlwd</code> argument.  Reported by Jurgen Kluge.
 </p>
 </li>
 <li> <p><code>svd(x)</code> and <code>eigen(x)</code> now behave analogously to
 <code>qr(x)</code> in accepting logical matrices <code>x</code>.
 </p>
 </li>
 <li><p> File &lsquo;<span class="file">NEWS</span>&rsquo; is now in UTF-8, and has a BOM (often
 invisible) on the first line, and <code>Emacs</code> local variables
 set for UTF-8 at the end.  <code>RShowDoc("NEWS")</code> should display
 this correctly, given suitable fonts.
 </p>
 </li>
 <li> <p><code>terms.formula(simplify = TRUE)</code> now does not deparse
 the LHS and so preserves non-standard responses such as <code>`a:
        b`</code> (requested by Sundar Dorai-Raj).
 </p>
 </li>
 <li><p> New function <code>news()</code> for building and querying <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> or
 package news information.
 </p>
 </li>
 <li> <p><code>z^n</code> for integer <code>n</code> and complex <code>z</code> is more
 accurate now if <code>|n| &lt;= 65536</code>.
 </p>
 </li>
 <li> <p><code>factor(NULL)</code> now returns the same as
 <code>factor(character(0))</code> instead of an error, and
 <code>table(NULL)</code> consequently does analogously.
 </p>
 </li>
 <li> <p><code>as.data.frame.vector()</code> (and its copies) is slightly faster
 by avoiding a copy if there are no names (following a
 suggestion of Tim Hesterberg).
 </p>
 </li>
 <li> <p><code>writeLines()</code>, <code>writeBin()</code> and
 <code>writeChar()</code> have a new argument <code>useBytes</code>.  If
 false, character strings with marked encodings are translated to
 the current locale (as before) but if true they are written
 byte-by-byte.
 </p>
 </li>
 <li> <p><code>iconv()</code> has a new argument <code>mark</code> which can be
 used (by experts) to suppress the declaration of encodings.
 </p>
 </li>
 <li> <p>&lsquo;<span class="file">DESCRIPTION</span>&rsquo; file's <code>LinkingTo</code> specs are now
 recognized as installation dependencies, and included in package
 management computations.
 </p>
 </li>
 <li><p> Standardized &lsquo;<span class="file">DESCRIPTION</span>&rsquo; file <code>License</code> specs
 are now available for package management computations.
 </p>
 </li>
 <li> <p><code>"\uxxxx"</code> and <code>"\Uxxxxxxxx"</code> escapes can now be
 parsed to a UTF-8 encoded string even in non-UTF-8 locales (this
 has been implemented on Windows since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.7.0).  The semantics
 have been changed slightly: a string containing such escapes is
 always stored in UTF-8 (and hence is suitable for portably
 including Unicode text in packages).
 </p>
 </li>
 <li><p> New <code>as.raw()</code> method for <code>"tclObj"</code> objects (wish
 of <a href="https://bugs.R-project.org/show_bug.cgi?id=13758">PR#13758</a>).
 </p>
 </li>
 <li> <p><code>Rd.sty</code> now makes a better job of setting email addresses,
 including using a monospaced font.
 </p>
 </li>
 <li> <p><code>textConnection()</code> gains an <code>encoding</code> argument
 to determine how input strings with marked encodings will be
 handled.
 </p>
 </li>
 <li> <p><code>R CMD Rd2pdf</code> is available as a shortcut for
 <code>R CMD Rd2dvi --pdf</code>.
 </p>
 </li>
 <li> <p><code>R CMD check</code> now checks links where a package is
 specified (<code>\link[pkg]{file}</code> or
 <code>\link[pkg:file]{topic}</code>), if the package is available.
 It notes if the package is not available, as in many cases this
 is an error in the link.
 </p>
 </li>
 <li> <p><code>identical()</code> gains three logical arguments, which
 allow for even more differentiation, notably <code>-0</code> and
 <code>0</code>.
 </p>
 </li>
 <li> <p><code>legend()</code> now can specify the <code>border</code> color of
 filled boxes, thanks to a patch from Frederic Schutz.
 </p>
 </li>
 <li><p> Indexing with a vector index to <code>[[ ]]</code> has now been
 extended to all recursive types.
 </p>
 </li>
 <li><p> Pairlists may now be assigned as elements of lists.  (Lists
 could always be created with pairlist elements, but <code>[[&lt;-</code>
 didn't support assigning them.)
 </p>
 </li>
 <li><p> The parser now supports C-preprocessor-like <code>#line</code>
 directives, so error messages and source references may refer to
 the original file rather than an intermediate one.
 </p>
 </li>
 <li><p> New functions <code>findLineNum()</code> and
 <code>setBreakpoint()</code> work with the source references to find
 the location of source lines and set breakpoints (using
 <code>trace()</code>) at those lines.
 </p>
 </li>
 <li><p> Namespace importing is more careful about warning on masked
 generics, thanks to a patch by Yohan Chalabi.
 </p>
 </li>
 <li> <p><code>detach()</code> now has an argument <code>character.only</code>
 with the same meaning as for <code>library()</code> or
 <code>require()</code>.
 </p>
 </li>
 <li> <p><code>available.packages()</code> gains a <code>filters</code>
 argument for specifying the filtering operations performed on the
 packages found in the repositories.  A new built-in
 <code>"license/FOSS"</code> filter only retains packages for which
 installation can proceed solely based on packages which can be
 verified as Free or Open Source Software (FOSS) employing the
 available license specifications.
 </p>
 </li>
 <li><p> In registering an S3 class by a call to
 <code>setOldClass()</code>, the data part (e.g., the object type)
 required for the class can be included as one of the superclasses
 in the Classes argument.
 </p>
 </li>
 <li><p> The argument <code>f</code> to <code>showMethods()</code> can be an
 expression evaluating to a generic function, allowing methods to
 be shown for non-exported generics and other nonstandard cases.
 </p>
 </li>
 <li> <p><code>sprintf()</code> now supports <code>%o</code> for octal
 conversions.
 </p>
 </li>
 <li><p> New function <code>Sys.readlink()</code> for information about symbolic
 links, including if a file is a symbolic link.
 </p>
 </li>
 <li><p> Package <span class="pkg">tools</span> has new functions
 <code>checkRdaFiles()</code> and <code>resaveRdaFiles()</code> to report on
 the format of &lsquo;<span class="file">.rda</span>&rsquo;/&lsquo;<span class="file">.RData</span>&rsquo; data files, and to
 re-save them in a different compressed format, including choosing
 the most compact format available.
 </p>
 <p>A new <code>INSTALL</code> option, <span class="option">--resave-data</span>, makes use
 of this.
 </p>
 </li>
 <li><p> File &lsquo;<span class="file">~/.R/config</span>&rsquo; is used in preference to
 &lsquo;<span class="file">~/.Rconfig</span>&rsquo;, and these are now documented in &lsquo;R
 Installation and Administration&rsquo;.
 </p>
 </li>
 <li><p> Logic operations with complex numbers now work, as they were always
 documented to, and as in S.
 </p>
 </li>
 <li> <p><code>arrows()</code> and <code>segments()</code> allow one of <code>x1</code>
 or <code>y1</code> to be omitted to simplify the specification of
 vertical or horizontal lines (suggestion of Tim Hesterberg).
 </p>
 </li>
 <li> <p><code>approxfun()</code> is faster by avoiding repeated <code>NA</code>
 checks (diagnosis and patch by Karline Soetaert &amp; Thomas
 Petzoldt).
 </p>
 </li>
 <li><p> There are the beginnings of a Nynorsk translation by Karl Ove
 Hufthammer.
 </p>
 </li>
 <li> <p><code>stripchart()</code> allows <code>par</code> <code>bg</code> to be
 passed in for the background colour for <code>pch = 21</code> (wish of
 <a href="https://bugs.R-project.org/show_bug.cgi?id=13984">PR#13984</a>).
 </p>
 </li>
 <li><p> New generic function <code>.DollarNames()</code> to enable class authors
 to customize completion after the <code>$</code> extractor.
 </p>
 </li>
 <li> <p><code>load()</code>, <code>save()</code>, <code>dput()</code> and
 <code>dump()</code> now open a not-yet-open connection in the
 appropriate mode (as other functions using connections directly
 already did).
 </p>
 </li></ul>


 <h4>REGULAR EXPRESSIONS</h4>


 <ul>
 <li><p> A different regular expression engine is used for basic and
 extended regexps and is also for approximate matching.  This is
 based on the TRE library of Ville Laurikari, a modified copy of
 which is included in the <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> sources.
 </p>
 <p>This is often faster, especially in a MBCS locale.
 </p>
 <p>Some known differences are that it is less tolerant of invalid
 inputs in MBCS locales, and in its interpretation of undefined
 (extended) regexps such as <code>"^*"</code>.  Also, the interpretation of
 ranges such as <code>[W-z]</code> in caseless matching is no longer to map
 the range to lower case.
 </p>
 <p>This engine may in future be used in &lsquo;literal&rsquo; mode for
 <code>fixed = TRUE</code>, and there is a compile-time option in
 <code>src/main/grep.c</code> to do so.
 </p>
 </li>
 <li><p> The use of repeated boundary regexps in <code>gsub()</code> and
 <code>gregexpr()</code> as warned about in the help page does not work
 in this engine (it did in the previous one since 2005).
 </p>
 </li>
 <li><p> Extended (and basic) regexps now support same set of options
 as for <code>fixed = TRUE</code> and <code>perl = TRUE</code>, including
 <code>useBytes</code> and support for UTF-8-encoded strings in
 non-UTF-8 locales.
 </p>
 </li>
 <li> <p><code>agrep()</code> now has full support for MBCS locales with a modest
 speed penalty.  This enables <code>help.search()</code> to use approximate
 matching character-wise rather than byte-wise.
 </p>
 </li>
 <li> <p><code>[g]sub</code> use a single-pass algorithm instead of
 matching twice and so is usually faster.
 </p>
 </li>
 <li><p> The <code>perl = TRUE</code> versions now work correctly in a
 non-UTF-8 MBCS locale, by translating the inputs to UTF-8.
 </p>
 </li>
 <li> <p><code>useBytes = TRUE</code> now inhibits the translation of inputs with
 marked encodings.
 </p>
 </li>
 <li> <p><code>strsplit()</code> gains a <code>useBytes</code> argument.
 </p>
 </li>
 <li><p> The algorithm used by <code>strsplit()</code> has been reordered
 to batch by elements of <code>split</code>: this can be much faster for
 <code>fixed = FALSE</code> (as multiple compilation of regexps is avoided).
 </p>
 </li>
 <li><p> The help pages, including <code>?regexp</code>, have been updated
 and should be consulted for details of the new implementations.
 </p>
 </li></ul>


 <h4>HELP &amp; Rd FILE CHANGES</h4>


 <ul>
 <li><p> A new dynamic HTML help system is used by default, and may be
 controlled using <code>tools::startDynamicHelp()</code>.  With this enabled,
 HTML help pages will be generated on request, resolving links
 by searching through the current <code>.libPaths()</code>.  The user
 may set options(<code>"help.ports"</code>) to control which IP port is used
 by the server.
 </p>
 </li>
 <li> <p><code>help.start()</code> no longer sets <code>options(htmlhelp =
        TRUE)</code> (it used to on Unix but not on Windows).  Nor does it on
 Unix reset the <code>"browser"</code> option if given an argument of
 that name.
 </p>
 <p>Arguments <code>update</code> and <code>remote</code> are now available
 on all platforms: the default is <code>update = FALSE</code> since the
 http server will update the package index at first use.
 </p>
 </li>
 <li> <p><code>help()</code> has a new argument <code>help_type</code> (with
 default set by the option of that name) to supersede arguments
 <code>offline</code>, <code>htmlhelp</code> and <code>chmhelp</code>
 (although for now they still work if <code>help_type</code> is
 unset).  There is a new type, <code>"PDF"</code> to allow offline PDF
 (rather than PostScript).
 </p>
 <p>A function <code>offline_help_helper()</code> will be used if this exists
 in the workspace or further down the search path, otherwise
 the function of that name in the <span class="pkg">utils</span> namespace is used.
 </p>
 </li>
 <li><p> Plain text help is now used as the fallback for HTML help (as
 it always was for Compiled HTML help on Windows).
 </p>
 </li>
 <li><p> It is possible to ask for static HTML pages to be prebuilt
 <em>via</em> the configure option <span class="option">--enable-prebuilt-html</span>.  This
 may be useful for those who wish to make HTML help available
 outside R, e.g. on a local web site.
 </p>
 </li>
 <li><p> An experimental tag <code>\Sexpr</code> has been added to Rd
 files, to evaluate expressions at build, install, or render time.
 Currently install time and render time evaluation are supported.
 </p>
 </li>
 <li><p> Tags <code>\if</code>, <code>\ifelse</code> and <code>\out</code> have been
 added to allow format-specific (or more general, using
 <code>\Sexpr</code>) conditional text in man pages.
 </p>
 </li>
 <li><p> The <code>parse_Rd()</code> parser has been made more tolerant of coding
 errors in Rd files:  now all syntax errors are reported as
 warnings, and an attempt is made to continue parsing.
 </p>
 </li>
 <li> <p><code>parse_Rd()</code> now has an argument <code>fragment</code>
 (default <code>FALSE</code>) to accept small fragments of Rd files (so
 that <code>\Sexpr</code> can output Rd code which is then parsed).
 </p>
 </li>
 <li> <p><code>parse_Rd()</code> now always converts its input to UTF-8.
 The <code>Rd2*</code> rendering functions have a new argument,
 <code>outputEncoding</code>, which controls how their output is
 encoded.
 </p>
 </li>
 <li> <p><code>parse_Rd()</code> no longer includes the newline as part of
 a <code>"%"</code>-style comment.
 </p>
 </li>
 <li><p> There have been various bug fixes and code reorganization in
 the Rd renderers <code>Rd2HTML</code>, <code>Rd2latex</code>, <code>Rd2txt</code>,
 and <code>Rd2ex</code>.
 </p>
 <p>All example files are now created with either ASCII or UTF-8
 encoding, and the encoding is only marked in the file if there
 is any non-UTF-8 code (previously it was marked if the help
 file had non-ASCII contents, possibly in other sections).
 </p>
 </li>
 <li> <p><code>print.Rd()</code> now adds necessary escape characters so that
 printing and re-parsing an Rd object should produce an
 equivalent object.
 </p>
 </li>
 <li> <p><code>parse_Rd()</code> was incorrectly handling multiple backslashes in
 <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> code strings, converting 4n+3 backslashes to 2n+1 instead
 of 2n+2.
 </p>
 </li>
 <li> <p><code>parse_Rd()</code> now recognizes the <code>\var</code> tag within
 a quoted string in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>-like text.
 </p>
 </li>
 <li> <p><code>parse_Rd()</code> now treats the argument of
 <code>\command</code> as LaTeX-like, rather than verbatim.
 </p>
 </li></ul>


 <h4>COMPRESSION</h4>


 <ul>
 <li><p> New function <code>untar()</code> to list or unpack tar archives,
 possibly compressed.  This uses either an external <code>tar</code>
 command or an internal implementation.
 </p>
 </li>
 <li><p> New function <code>tar()</code> to create (possibly compressed)
 tar archives.
 </p>
 </li>
 <li><p> New functions <code>memCompress()</code> and
 <code>memDecompress()</code> for in-memory compression and
 decompression.
 </p>
 </li>
 <li> <p><code>bzfile()</code> has a <code>compress</code> argument to select
 the amount of effort put into compression when writing.
 </p>
 </li>
 <li><p> New function <code>xzfile()</code> for use with
 <code>xz</code>-compressed files.  (This can also read files
 compressed by some versions of <code>lzma</code>.)
 </p>
 </li>
 <li> <p><code>gzfile()</code> looks at the file header and so can now also
 read <code>bzip2</code>-ed files and <code>xz</code>-compressed files.
 </p>
 </li>
 <li><p> There are the new options of <code>save(compress = "bzip2")</code>
 and <code>"xz"</code> to use <code>bzip2</code> or <code>xz</code>
 compression (which will be slower, but can give substantially
 smaller files).  Argument <code>compression_level</code> gives finer
 control over the space/time tradeoffs.
 </p>
 <p><code>load()</code> can read such saves (but only as from this version of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>).
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code>/<code>check</code> and
 <code>tools::writePACKAGES</code> accept a wider range of compressed
 tar archives.  Precisely how wide depends on the capabilities of
 the host system's <code>tar</code> command: they almost always
 include <code>.tar.bz2</code> archives, and with modern versions of
 <code>tar</code> other forms of compression such as <code>lzma</code>
 and <code>xz</code>, and arbitrary extensions.
 </p>
 </li>
 <li> <p><code>R CMD INSTALL</code> has a new option
 <span class="option">--data-compress</span> to control the compression used when
 lazy-loading data.  New possibilities are
 <span class="option">--data-compress=bzip2</span> which will give ca 15% better
 compression at the expense of slower installation times, and
 <span class="option">--data-compress=xz</span>, often giving even better compression
 on large datasets at the expense of much longer installation
 times.  (The latter is used for the recommended packages: it is
 particularly effective for <a href="https://CRAN.R-project.org/package=survival"><span class="pkg">survival</span></a>.)
 </p>
 </li>
 <li> <p><code>file()</code> for open = <code>""</code>, <code>"r"</code> or
 <code>"rt"</code> will automagically detect compressed files (from
 <code>gzip</code>, <code>bzip2</code> or <code>xz</code>).  This means
 that compressed files can be specified by file name (rather than
 <em>via</em> a <code>gzfile()</code> connection) to <code>read.table()</code>,
 <code>readlines()</code>, <code>scan()</code> and so on.
 </p>
 </li>
 <li> <p><code>data()</code> can handle compressed text files with extensions
 &lsquo;<span class="file">.{txt,tab,csv}.{gz,bz2,xz}</span>&rsquo; .
 </p>
 </li></ul>


 <h4>DEPRECATED &amp; DEFUNCT</h4>


 <ul>
 <li> <p><code>png(type="cairo1")</code> is defunct: the value is no longer
 recognized.
 </p>
 </li>
 <li> <p><code>tools::Rd_parse()</code> is defunct (as this version of <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span>
 uses only Rd version 2).
 </p>
 </li>
 <li><p> Use of file &lsquo;<span class="file">~/.Rconf</span>&rsquo; (which was deprecated in favour
 of &lsquo;<span class="file">~/.Rconfig</span>&rsquo; in 2004) has finally been removed.
 </p>
 </li>
 <li><p> Bundles of packages are deprecated.  See &lsquo;Writing R
 Extensions&rsquo; for the steps needed to unbundle a bundle.
 </p>
 </li>
 <li> <p><code>help()</code> arguments <code>offline</code>, <code>htmlhelp</code> and
 <code>chmhelp</code> are deprecated in favour of <code>help_type</code>.
 </p>
 </li>
 <li> <p><code>clearNames()</code> (in package <span class="pkg">stats</span>) is deprecated
 for <code>unname()</code>.
 </p>
 </li>
 <li><p> Basic regular expressions (<code>extended = FALSE</code>) are
 deprecated in <code>strsplit</code>, <code>grep</code> and friends.  There is
 a precise POSIX standard for them, but it is not what recent RE
 engines implement, and it seems that in almost all cases package
 authors intended <code>fixed = TRUE</code> when using <code>extended =
        FALSE</code>.
 </p>
 </li>
 <li> <p><code>methods::trySilent()</code> is deprecated in favour of
 <code>try(silent=TRUE)</code> or &ndash; more efficiently and flexibly &ndash;
 something like <code>tryCatch(error = function(e) e)</code>.
 </p>
 </li>
 <li> <p><code>index.search()</code> is deprecated: there are no longer
 directories of types other than <code>help</code>.
 </p>
 </li></ul>


 <h4>INSTALLATION</h4>


 <ul>
 <li><p> cairo &gt;= 1.2 is now required (1.2.0 was released in July 2006)
 for cairo-based graphics devices (which remain optional).
 </p>
 </li>
 <li><p> A suitable <code>iconv()</code> is now required: support for
 configure option <span class="option">--without-iconv</span> has been withdrawn (it
 was deprecated in <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.5.0).
 </p>
 </li>
 <li><p> Perl is no longer &lsquo;essential&rsquo;.  <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> can be built
 without it, but scripts <code>R CMD build</code>, <code>check</code>,
 <code>Rprof</code> and <code>Sd2d</code> currently require it.
 </p>
 </li>
 <li><p> A system <code>glob</code> function is now essential (a working
 <code>Sys.glob()</code> has been assumed since <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.9.0 at least).
 </p>
 </li>
 <li><p> C99 support for MBCS is now required, and configure option
 <span class="option">--disable-mbcs</span> has been withdrawn.
 </p>
 </li>
 <li><p> Having a version of <code>tar</code> capable of automagically
 detecting compressed archives is useful for
 <code>utils::untar()</code>, and so <code>gtar</code> (a common name for
 GNU <code>tar</code>) is preferred to <code>tar</code>: set environment
 variable <span class="env">TAR</span> to specify a particular <code>tar</code> command.
 </p>
 </li></ul>


 <h4>INTERNATIONALIZATION</h4>


 <ul>
 <li><p> There is some makefile support for adding/updating translations
 in packages: see <code>po/README</code> and &lsquo;Writing R Extensions&rsquo;.
 </p>
 <p>There is support for the use of <code>dngettext</code> for C-level
 translations in packages: see &lsquo;Writing R Extensions&rsquo;.
 </p>
 </li></ul>


 <h4>BUG FIXES</h4>


 <ul>
 <li><p> Assigning an extra 0-length column to a data frame by
 <code>DF[, "foo"] &lt;- value</code> now works in most cases (by filling
 with <code>NA</code>s) or fails.  (It used to give a corrupt data
 frame.)
 </p>
 </li>
 <li> <p><code>validObject()</code> avoids an error during evaluation in the case
 of various incorrect slot definitions.
 </p>
 </li>
 <li> <p><code>n:m</code> now returns a result of type <code>"integer"</code> in
 a few more boundary cases.
 </p>
 </li>
 <li><p> The <code>zap.ind</code> argument to <code>printCoefmat()</code> did
 not usually work as other code attempted to ensure that non-zero
 values had a non-zero representation.
 </p>
 </li>
 <li> <p><code>printCoefmat()</code> formatted groups of columns together,
 not just the <code>cs.ind</code> group but also the <code>zap.ind</code>
 group and a residual group.  It now formats all columns except
 the <code>cs.ind</code> group separately (and zaps the <code>zap.ind</code>
 group column-by-column).  The main effect will be see in the
 output from <code>print.anova()</code>, as this grouped SS-like columns
 in the <code>zap.ind</code> group.
 </p>
 </li>
 <li> <p><code>R_ReplDLLinit()</code> initializes the top-level jump so
 that some embedded applications on Windows no longer crash on
 error.
 </p>
 </li>
 <li> <p><code>identical()</code> failed to take the encoding of character
 strings into account, so identical byte patterns are not
 necessarily identical strings, and similarly Latin-1 and UTF-8
 versions of the same string differ in byte pattern.
 </p>
 </li>
 <li> <p><code>methods(f)</code> used to warn unnecessarily for an S4
 generic <code>f</code> which had been created based on an existing S3
 generic.
 </p>
 </li>
 <li><p> The check for consistent ordering of superclasses was not
 ignoring all conditional relations (the symptom was usually
 spurious warnings for classes extending <code>"array"</code>).
 </p>
 </li>
 <li><p> Trying to assign into a raw vector with an index vector
 containing <code>NA</code>s could cause a segfault.  Reported by Hervé
 Pagès.
 </p>
 </li>
 <li> <p><code>Rscript</code> could segfault if (by user error) its
 filename argument was missing.  Reported by Martin Morgan.
 </p>
 </li>
 <li> <p><code>getAnywhere()</code> (and functions that use it, including argument
 completion in the console) did not handle special built-in
 functions.  Reported by Romain Francois.
 </p>
 </li>
 <li> <p><code>order()</code> was missing a <code>PROTECT()</code> call and so
 could segfault when called on character data under certain (rare)
 circumstances involving marked non-native encodings.
 </p>
 </li>
 <li> <p><code>prettyNum(z, drop0trailing=TRUE)</code> did not work
 correctly when <code>z</code> was a complex vector.  Consequently,
 <code>str(z, ...)</code> also did not.  (<a href="https://bugs.R-project.org/show_bug.cgi?id=13985">PR#13985</a>)
 </p>
 </li>
 <li> <p><code>make distclean</code> removed too many files in
 <code>etc/</code> if builddir = srcdir.
 </p>
 </li>
 <li> <p><code>R CMD</code> replaced <code>TEXINPUTS</code> rather than
 appending to it (as documented and intended).
 </p>
 </li>
 <li> <p><code>help.start()</code> no longer fails on unix when
 <code>"browser"</code> is a function.
 </p>
 </li>
 <li> <p><code>pbeta(x, ..., log.p = TRUE)</code> is sometimes more accurate,
 e.g., for very small <code>x</code>.
 </p>
 </li>
 <li><p> Unserializing a pre-2.8 workspace containing pure ASCII
 character objects with a Latin-1 or UTF-8 encoding would corrupt
 the <code>CHARSXP</code> cache.
 </p>
 </li></ul>


 <h3>OLDER NEWS</h3>

 <p>News for <span style="font-family: Courier New, Courier; color: #666666;"><b>R</b></span> 2.9.2 patched and earlier can be found in files
 such as &lsquo;<span class="file">NEWS.2</span>&rsquo; in the &lsquo;<span class="file">doc</span>&rsquo; directory.
 </p>


 </div></body></html>