aaron.ballman added a comment. In D64671#1644280 <https://reviews.llvm.org/D64671#1644280>, @jpakkane wrote:
> I used stdint to replicate a real world use case as I'd imagine those types > would match this search quite heavily. The problem is that this uses the system header search paths and so the testing system has no control over *what* is found from bot to bot. That is why we structure almost all of our tests to not use system includes. > The tests already have one test for a typedeffed integer and one that is > defined with a macro. If those are deemed sufficient, the stdint type can be > removed altogether. If the stdint types are defined via some other mechanism > such as compiler intrinsics, there would be no test for that. I do not know > if that is the case with current libcs. I don't see a need for int32_t specifically in this test. > I can update the MR as requested once someone with enough stripes makes the > policy decision. I'm not certain that we've written this instruction down so much as let new contributors know when they stumble into it. It would probably be a good idea for us to get some testing documentation written down at some point. Despite that, our policy has been to not use `#include` in test cases unless the included file is relative to the test. CHANGES SINCE LAST ACTION https://reviews.llvm.org/D64671/new/ https://reviews.llvm.org/D64671 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
