Bruno Haible <[email protected]> writes:
>> Posting on bug-gnulib since this test failure seems to be caused by
>> parse-datetime despite being caught by tests/date/date-debug.sh in
>> Coreutils.
>>
>> On NetBSD 10.0 (cfarm428.cfarm.net) it fails with the following:
>
> This is a NetBSD 10.0/arm64 machine. What about NetBSD 10.0/x86_64?
> If you tell me the exact "How to reproduce" recipe, I could try it on
> my NetBSD 10.0/x86_64 VM.
I don't think this is needed since I think this has already been
confirmed to fail on NetBSD 10.0/x86_64 from your CI. There is a patch
that is applied marking it XFAIL on that platform [1].
But I built a tarball on my host machine:
$ git clone https://git.savannah.gnu.org/git/coreutils.git
$ ./bootstrap && ./configure && make && make dist
$ scp coreutils-9.7.43-2f6a.tar.xz cfarm428.cfarm.net:~/
Then on the NetBSD machine:
$ tar -xf coreutils-9.7.43-2f6a.tar.xz
$ cd coreutils-9.7.43-2f6a
$ ./configure && make && make check SHELL=bash
> PS: What I'm doing here is to follow the "get the entire picture first"
> approach, before even digging into the debugger and the logs.
Makes sense. I wrote the investigation but not how I got there. :)
Collin
P.S. Not sure if SHELL=bash is needed for Coreutils, but I use it since
NetBSD '/bin/sh' does not support sub-second resolution for it's 'test'
built-in. I submitted a patch for it, but it seems to have gotten lost
[2]. I fixed the same issue for OpenBSD 7.7 [3] after it caused
Autoconf's tests to fail [4].
[1]
https://github.com/coreutils/ci-check/blob/f25dac6256d14899c6c8cd40e7ead4c784d25cb5/patches/expected-failures.patch
[2] https://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=59210
[3]
https://github.com/openbsd/src/commit/873aa13239043fac57d01b96f0c4f18be6abe5eb
[4] https://lists.gnu.org/archive/html/autoconf-patches/2025-03/msg00001.html
