[PATCH] D77986: [lit] Move llvm-test-suite result codes into llvm/lit

Julian Lettner via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Apr 14 11:50:41 PDT 2020


yln added a comment.

In D77986#1980779 <https://reviews.llvm.org/D77986#1980779>, @jdenny wrote:

> > I don't think we should "rename" it in the test-suite as it is not "unresolved" (whatever that then means) but we just failed to produce an executable so the compile stage failed not the run stage (which would be FAIL).
>
> Based on that description, I agree with @jdoerfert that UNRESOLVED or SKIPPED doesn't make sense for NOEXE.  NOEXE sounds like a special kind of FAIL.


I mixed this up as well.  Thanks for explaining @jdoerfert!

> I don't have enough experience with lnt to know why it's worthwhile to distinguish NOEXE from FAIL, but it's not a new distinction in lnt.  If it's worthwhile there, maybe it will be worthwhile in other lit-based test suites too.

This is the question then.  Are there plans (or a desire) to adopt this notion in other lit tests as well and should this become a supported lit feature?  Is not having these categories in "vanilla" lit blocking improvements in lnt?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D77986/new/

https://reviews.llvm.org/D77986





More information about the llvm-commits mailing list