On Tue, 12 Aug 2025, Takashi Yano via Cygwin-apps wrote:

> On Sun, 10 Aug 2025 16:55:56 -0700 (PDT)
> Jeremy Drake wrote:
> > On Tue, 22 Jul 2025, Jeremy Drake via Cygwin-apps wrote:
> >
> > > On Sun, 15 Jun 2025, Jeremy Drake via Cygwin-apps wrote:
> > >
> > > > I pushed update to 20.1.7 and synced patches with upstream, current
> > > > test results:
> > > > llvm:
> > > > Total Discovered Tests: 65227
> > > >   Skipped          :    52 (0.08%)
> > > >   Unsupported      :  2197 (3.37%)
> > > >   Passed           : 62757 (96.21%)
> > > >   Expectedly Failed:   193 (0.30%)
> > > >   Failed           :    28 (0.04%)
> > > >
> > > > clang:
> > > > Total Discovered Tests: 46308
> > > >   Skipped          :     8 (0.02%)
> > > >   Unsupported      :   181 (0.39%)
> > > >   Passed           : 46069 (99.48%)
> > > >   Expectedly Failed:    31 (0.07%)
> > > >   Failed           :    19 (0.04%)
> > > >
> > > > lld:
> > > > Total Discovered Tests: 1066
> > > >   Unsupported:   10 (0.94%)
> > > >   Passed     : 1056 (99.06%)
> >
> > I went back to the "playground" branch (20.1.8), and backported more
> > (mostly test) fixes.  I removed the clang patch to call the linker
> > directly there, it was out of date, and probably makes more sense to
> > continue to develop against 21.1.
>
> The latest clang patch to call the linker directly cannot be applied
> 20.1.8 cleanly due to some changes in 21.1.
>
> > Current test results for 20.1.8:
> >
> > llvm:
> > Total Discovered Tests: 65233
> >   Skipped          :    55 (0.08%)
> >   Unsupported      :  2200 (3.37%)
> >   Passed           : 62769 (96.22%)
> >   Expectedly Failed:   194 (0.30%)
> >   Failed           :    15 (0.02%)
> >
> > clang:
> > Total Discovered Tests: 46308
> >   Skipped          :     8 (0.02%)
> >   Unsupported      :   192 (0.41%)
> >   Passed           : 46068 (99.48%)
> >   Expectedly Failed:    33 (0.07%)
> >   Failed           :     7 (0.02%)
>
> Great! Did you mark some tests as 'unsupported' for clang?

Yes, essentially.  There are a number of tests that are unsupported or
xfail on "system-windows".  I added a "system-cygwin" and marked those
tests that were failing on cygwin but already marked to exclude them on
windows to exclude them on cygwin in the same way.  I didn't extend
system-windows to include cygwin because there are also a number of tests
that don't work on windows that do work on cygwin.

>
> > Do you think it makes sense to deploy 20.1.8 to master soon and do at
> > least a test release?
>
> Yes I do! Will you do that?

Now there's one more test fix pending that apparently will fix the
ProfileData tests on Cygwin (these are 4 test failures on llvm).  This is
a fix to the test though, not the code being tested, so it shouldn't be an
issue.

Right now I have a GitHub Actions artifact from building the bootstrapping
of llvm and clang (from
https://github.com/jeremyd2019/cygwin-rust-bootstrap/actions/runs/16867148865).

This is a 11.5 hour workflow (split into 4 jobs), and the resulting
artifact is 1.64GB.  I imagine the best way to deploy this would be to
download locally and sftp.  After that, we should be able to use the last
version of clang to build the next version and avoid the bootstrapping
issue, and use scallywag.  Hopefully.  Though it sounds like windows-2025
images will be made windows-latest next month, and that they will have
significantly less available disk space
(https://github.com/actions/runner-images/issues/12609) and worse
performance (https://github.com/actions/runner-images/issues/12647).


I've uploaded that now

Reply via email to