This is an archive of the discontinued LLVM Phabricator instance.

[AIX] Adjust support of format function tests
ClosedPublic

Authored by Jake-Egan on Apr 24 2023, 10:09 AM.

Details

Summary

escaped_output.unicode.pass.cpp is failing only on 32-bit AIX. The rest are passing.

Diff Detail

Event Timeline

Jake-Egan created this revision.Apr 24 2023, 10:09 AM
Herald added a project: Restricted Project. · View Herald TranscriptApr 24 2023, 10:09 AM
Jake-Egan requested review of this revision.Apr 24 2023, 10:09 AM
Herald added a project: Restricted Project. · View Herald TranscriptApr 24 2023, 10:09 AM
Herald added a reviewer: Restricted Project. · View Herald Transcript
Jake-Egan edited the summary of this revision. (Show Details)Apr 24 2023, 10:11 AM
Jake-Egan added reviewers: daltenty, nicolerabjohn.
Jake-Egan abandoned this revision.Apr 24 2023, 1:03 PM

I see - I was testing on 64-bit, but this test only fails on 32-bit AIX because _LIBCPP_SHORT_WCHAR gets defined.

Jake-Egan reclaimed this revision.Apr 24 2023, 6:24 PM
Jake-Egan planned changes to this revision.
Jake-Egan retitled this revision from [AIX] Remove UNSUPPORTED from format function tests to [AIX] Adjust support of format function tests.
Jake-Egan edited the summary of this revision. (Show Details)
Mordante accepted this revision.Apr 27 2023, 10:03 AM
Mordante added a subscriber: Mordante.

LGTM when the CI passes. Since I don't have access to a system with a 16-bit wchar_t it would be great if you could look at fixing the tests too.

This revision is now accepted and ready to land.Apr 27 2023, 10:03 AM
Jake-Egan updated this revision to Diff 517896.Apr 28 2023, 6:40 AM
Jake-Egan updated this revision to Diff 518427.May 1 2023, 6:15 AM
  • Trigger CI

Since I don't have access to a system with a 16-bit wchar_t it would be great if you could look at fixing the tests too.

@Mordante The test fails at the assert(false); line after calling test_ill_formed_utf16(). I haven't looked into the test in depth, so forgive me if it's obvious, but the other functions test_ill_formed_utf8() and test_ill_formed_utf32() don't assert false, so it looks like this was a mistake.

Jake-Egan updated this revision to Diff 518699.May 2 2023, 6:18 AM
  • Trigger CI
Jake-Egan updated this revision to Diff 519552.May 4 2023, 10:15 AM
  • Trigger CI

I'm still happy with this patch.

Since I don't have access to a system with a 16-bit wchar_t it would be great if you could look at fixing the tests too.

@Mordante The test fails at the assert(false); line after calling test_ill_formed_utf16(). I haven't looked into the test in depth, so forgive me if it's obvious, but the other functions test_ill_formed_utf8() and test_ill_formed_utf32() don't assert false, so it looks like this was a mistake.

At some point I will look into this. The main issue is that the CI doesn't give nice output when things fails for wide strings. I probably write an wide to narrow string transcoder so the CI has readable output. (But this is a bit low on my priority list.)

Jake-Egan updated this revision to Diff 520346.May 8 2023, 6:31 AM
  • Trigger CI
This revision was automatically updated to reflect the committed changes.