framework/tests/value-ignore-range-overflow/TransferSub.java

import org.checkerframework.common.value.qual.*;

public class TransferSub {

  void test() {
    // zero, one, and two
    int a = 1;

    @IntRange(from = 0) int b = a - 1;
    // :: error: (assignment)
    @IntRange(from = 1) int c = a - 1;
    @IntRange(from = -1) int d = a - 2;

    // :: error: (assignment)
    @IntRange(from = 0) int e = a - 2;

    @IntRange(from = -1) int f = b - 1;
    // :: error: (assignment)
    @IntRange(from = 0) int g = b - 1;

    // :: error: (assignment)
    @IntRange(from = -1) int h = f - 1;

    @IntRange(from = -1) int i = f - 0;
    @IntRange(from = 0) int j = b - 0;
    @IntRange(from = 1) int k = a - 0;

    // :: error: (assignment)
    @IntRange(from = 1) int l = j - 0;
    // :: error: (assignment)
    @IntRange(from = 0) int m = i - 0;

    // :: error: (assignment)
    @IntRange(from = 1) int n = a - k;
    // this would be an error if the values of b and j (both zero) weren't known at compile time
    @IntRange(from = 0) int o = b - j;
    /* i and d both have compile time value -1, so this is legal.
    The general case of GTEN1 - GTEN1 is not, though. */
    @IntRange(from = -1) int p = i - d;

    // decrements

    // :: error: (unary.decrement)
    // :: error: (assignment)
    @IntRange(from = 1) int q = --k; // k = 0

    // :: error: (unary.decrement)
    @IntRange(from = 0) int r = k--; // after this k = -1

    int k1 = 0;
    @IntRange(from = 0) int s = k1--;

    // :: error: (assignment)
    @IntRange(from = 0) int s1 = k1;

    k1 = 1;
    @IntRange(from = 0) int t = --k1;

    k1 = 1;
    // :: error: (assignment)
    @IntRange(from = 1) int t1 = --k1;

    int u1 = -1;
    @IntRange(from = -1) int x = u1--;
    // :: error: (assignment)
    @IntRange(from = -1) int x1 = u1;
  }
}