Re: [PATCH] ci: add Cirrus-CI config for FreeBSD CI

[Ed Maste <emaste@freebsd.org>]

On Mon, 25 Nov 2019 at 19:44, brian m. carlson
<sandals@crustytoothpaste.net> wrote:
>
> On 2019-11-25 at 20:37:40, Ed Maste wrote:
> > From: Ed Maste <emaste@freebsd.org>
> >
> > Cirrus-CI is relatively unique among hosted CI providers in supporting
> > FreeBSD (in addition to Linux, Windows, and macOS).  Add a Cirrus-CI
> > config to facilitate building and testing on FreeBSD.
> >
> > Signed-off-by: Ed Maste <emaste@freebsd.org>
>
> I'm all for automated testing on FreeBSD, but we would need someone to
> triage and address any failures reasonably quickly.  Is that something
> you'd be okay with doing, or is there someone else who would be okay
> with doing that?

We're currently experimenting with a migration of the FreeBSD repo
from Subversion to git, so long term there will be many with a vested
interest in triaging and addressing failures. In the near term though
I'd be able to take this on.

> > 6 i18n tests are currently failing and need investigation.
> > Example result: https://cirrus-ci.com/task/5394512637067264
> > Test log: https://api.cirrus-ci.com/v1/task/5394512637067264/logs/test.log
>
> Could we fix these issues first so we don't have CI suddenly start
> failing?

Indeed, that makes sense. I think the failures may be an issue with
the test though; here's the first failure:

expecting success of 4210.6 '-c grep.patternType=fixed log --grep does not find
non-reencoded values (latin1 + locale)':
                cat >expect <<-\EOF &&
                latin1
                utf8
                EOF
                LC_ALL="is_IS.UTF-8" git -c grep.patternType=fixed log
--encoding=ISO-8859-1 --format=%s --grep="_" >actual &&
                test_cmp expect actual

fatal: command line, '_': illegal byte sequence

This was added in 4e2443b1813 with this note in the commit message:

| It's possible that this
| test breaks the "basic" and "extended" backends on some systems that
| are more anal than glibc about the encoding of locale issues with
| POSIX functions that I can remember, but PCRE is more careful about
| the validation.

I've CC'd Ævar Arnfjörð Bjarmason for advice on this.