Skip to content

Conversation

@spcfox
Copy link
Contributor

@spcfox spcfox commented Aug 18, 2025

Description

Often, when changing compilers, differences in machine names are detected in tests. I think it would be useful to apply clean_names to all tests by default.

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

@@ -1 +1 @@
ERROR: Unhandled input for Prelude.Num.case block in mod at Prelude.Num:94:3--96:44
ERROR: Unhandled input for Prelude.Num.case block in mod at Prelude.Num:1
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we should remove the location in the names of implementations. This will make the tests resistant to changes in Prelude.

@spcfox
Copy link
Contributor Author

spcfox commented Aug 18, 2025

Perhaps we should only clean locators after (implementation | case block | with block) at?

@dunhamsteve
Copy link
Collaborator

dunhamsteve commented Aug 18, 2025

If we wanted to be precise we could do after (implementation | case block | with block) at but also after ^LOG.

My only reservation with what you have is that we might loose tests for locations of errors (e.g. sometimes unification errors will name a and put a location after it), but we could choose to not use clean_names in tests where it is an issue.

@spcfox spcfox marked this pull request as draft August 19, 2025 11:33
@spcfox spcfox marked this pull request as ready for review August 19, 2025 12:00
@spcfox
Copy link
Contributor Author

spcfox commented Aug 20, 2025

I think it shouldn't affect locations in errors now

@spcfox
Copy link
Contributor Author

spcfox commented Aug 20, 2025

Windows hangs every time in this PR. Maybe clean_names is hanging there?

@dunhamsteve
Copy link
Collaborator

dunhamsteve commented Aug 20, 2025

That could be, it looks like the while loop expects the replacement to leave something that doesn't match the original regex.

We need a termination checker for AWK. :)

@spcfox
Copy link
Contributor Author

spcfox commented Aug 25, 2025

Windows hangs every time in this PR. Maybe clean_names is hanging there?

I've made many runs. It seems like it's just bad luck

UPD: Or not. It doesn't always hang, but it seems to happen more often than usual.

@spcfox spcfox requested a review from gallais August 25, 2025 22:49
@spcfox spcfox force-pushed the clean-tests branch 3 times, most recently from 53dee91 to 5e66ce7 Compare September 1, 2025 23:19
@spcfox
Copy link
Contributor Author

spcfox commented Sep 9, 2025

Windows hangs every time in this PR. Maybe clean_names is hanging there?

I've made many runs. It seems like it's just bad luck

UPD: Or not. It doesn't always hang, but it seems to happen more often than usual.

The reason is the channels009 test. There were problems with it before. I suggest disabling it temporarily. PR #3617 is working on fixing it.

@gallais
Copy link
Member

gallais commented Oct 23, 2025

Happy to merge this whenever it's ready to go.

@gallais gallais merged commit 1a5fbb2 into idris-lang:main Oct 24, 2025
22 checks passed
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.

3 participants