| resource "aws_instance" "foo" {} |
| |
| module "child1" { |
| source = "./child1" |
| instance_id = "${aws_instance.foo.id}" |
| } |
| |
| module "child2" { |
| source = "./child2" |
| } |
| |
| output "child1_id" { |
| value = "${module.child1.instance_id}" |
| } |
| |
| output "child1_given_id" { |
| value = "${module.child1.given_instance_id}" |
| } |
| |
| output "child2_id" { |
| # This should get updated even though we're targeting specifically |
| # module.child2, because outputs are implicitly targeted when their |
| # dependencies are. |
| value = "${module.child2.instance_id}" |
| } |
| |
| output "all_ids" { |
| # Here we are intentionally referencing values covering three different scenarios: |
| # - not targeted and not already in state |
| # - not targeted and already in state |
| # - targeted |
| # This is important because this output must appear in the graph after |
| # target filtering in case the targeted node changes its value, but we must |
| # therefore silently ignore the failure that results from trying to |
| # interpolate the un-targeted, not-in-state node. |
| value = "${aws_instance.foo.id} ${module.child1.instance_id} ${module.child2.instance_id}" |
| } |
