Re: libgit2

On Sun, 25 May 2025, Jon Turney wrote:

> On 24/05/2025 19:27, Jeremy Drake via Cygwin-apps wrote:
> > On Sat, 24 May 2025, ASSI via Cygwin-apps wrote:
> >
> > > Jeremy Drake via Cygwin-apps writes:
> > > > I made a little attempt to update libgit2 based on what MSYS2 did, since
>
> Thanks for looking into this.
>
> I'm afraid our process for dealing with drive-up cygport updates is not very
> reliable (since it relies on someone with appropriate privs noting the update
> and applying it), but we should be able to do that in this case :)
>
> > Attaching updated cygport, scallywag run at
> > https://github.com/cygwin/scallywag/actions/runs/15229587565/job/42835871879
> >
> > 83% tests passed, 2 tests failed out of 12
> > Total Test time (real) = 1182.89 sec
> > The following tests FAILED:
> > 	  1 - offline (Failed)
> > 	 12 - util (Failed)
> > Errors while running CTest
> >
> > Do I need to probably somehow skip the remaining failing tests (they seem
> > generally to be related to Cygwin unicode quirks), or just put
> > SCALLYWAG="notest" in the cygport?
>
> I think marking as notest (perhaps with a comment memorializing the specifcs)
> is fine.
>
> I notice that the tests pass (with one hang) when I try against the current
> version we have (0.25.1), but due to the ridiculous amount of time that has
> passed, ascribing that to new tests or upstream regressions is more work than
> it should be...


Since Cygwin 3.6.2 was released I pushed another update removing the
'extensions' workaround, and adding ` || true` to the call to ctest.

Should I add a big comment documenting the test failures, or a separate
file?


12:   2) Failure:
12: path::core::validate_current_user_ownership
[/usr/src/debug/libgit2-1.9.0-1/tests/util/path/core.c:764]
12:   is_cur != (geteuid() == 0)
12:   1 != 0
12:

I looked into this one, and is_cur is true if `/` is owned by the current
user.  This test expects that `/` is owned by uid 0.