Skip to content

Conversation

@spcfox
Copy link
Contributor

@spcfox spcfox commented Nov 26, 2025

Description

The tests idris/basic/case001 and codegen/enum fail when the compiler is built using Racket. We could have caught this earlier if ubuntu-self-host-racket were being run in CI. It looks like it was disabled many years ago, and since it is currently running without errors, I propose re-enabling it. I've set the timeout to 60 minutes because Racket runs more slowly than Chez (the other tests use 45 minutes).

Some additional fixes to the tests:

  1. Tests for specific backends have been moved to the corresponding directories.
  2. Removed the codegen specification from run for tests located in backend directories, as their code generator is specified in tests/Main.idr. Specifying the code generator in run makes a test less portable, because it causes the test to ignore the code generator set in Main.idr. For example, if such a test is copied from chez to racket, it would still use Chez. Therefore, I suggest avoiding duplication of the code generator specification in run.
  3. Added debug output in codegen/fix3515. Issue Miscompilation of Nat literals #3515 is not backend-specific, so it’s useful to check that compexpr has not changed.
  4. Removed unused argument of the findTests function in Golden.

Self-check

  • This is my first time contributing, I've carefully read CONTRIBUTING.md
    and I've updated CONTRIBUTORS.md with my name.
  • If this is a fix, user-facing change, a compiler change, or a new paper
    implementation, I have updated CHANGELOG_NEXT.md

@spcfox spcfox changed the title [ ci, test ] Test compiler builded with Racket and some fixed in tests [ ci, test ] Test compiler builded with Racket and some fixes in tests Nov 26, 2025
@buzden
Copy link
Collaborator

buzden commented Nov 27, 2025

I think this re-enabling is reasonable, despite bigger time values, because as far as I remember, we removed those tests prior we had "quick-check" job and prior to decision to stop jobs of the same branch on new commits. None of this arguments are relevant now.

Also, putting things in order in backend-related stuff seems to good pretty good.

@spcfox
Copy link
Contributor Author

spcfox commented Dec 6, 2025

In #3691 I added a test case003. This test is not for case but for casetree, so I moved it

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants