Re: [PATCH] gitlab-ci: add smoke test for fuzzers

From: Patrick Steinhardt <ps@pks.im>
Date: Apr 30, 2024, 04:30 UTC
Message-ID: <ZjBz_V31u_9CfXDn@tanuki>
In-Reply-To: <xmqqfrv4yyb6.fsf@gitster.g>

On Mon, Apr 29, 2024 at 08:37:49AM -0700, Junio C Hamano wrote:

Show 29 quoted lines

> Patrick Steinhardt <ps@pks.im> writes:
> 
> > On Mon, Apr 29, 2024 at 08:13:23AM +0200, Patrick Steinhardt wrote:
> >> Our GitLab CI setup has a test gap where the fuzzers aren't exercised at
> >> all. Add a smoke test, similar to the one we have in GitHub Workflows.
> >> 
> >> Signed-off-by: Patrick Steinhardt <ps@pks.im>
> >> ---
> >> 
> >> As identified by Junio in <xmqqwmoi31aw.fsf@gitster.g>.
> >> 
> >> Patrick
> >
> > I forgot to add the link to a successful run of this job:
> > https://gitlab.com/gitlab-org/git/-/jobs/6735705569
> 
> Thanks.  I wonder if we can somehow automate a change like this.
> 
> Seeing how simple this fix has become thanks to the use of
> before_script/script pair that merely point at ci/*.sh scripts,
> perhaps we have already extracted enough commonalities as a set of
> shell scripts in ci/ hierarchy.  I wonder if we can have a common
> "source" that is "compiled" into .gitlab-ci.yml and its counterpart
> for GitHub Actions?
> 
> Or perhaps a linter that can say things like "ah, you are adding
> this new test to one, but not touching the other, shouldn't you?",
> and "you are tweaking this existing test in one, but shouldn't you
> be doing the same to the other?"

We probably could, yeah. The question is whether it would really be worth it in the end. GitLab CI is still a relatively new addition, and thus it needs to catch up with what GitHub Workflows has. But once that is done I don't expect there to be a ton of changes to the CI setup, and the few new additions that we gain once in a while should be relatively easy to spot during review.

So if anybody is up for it then I'm happy to review that. But I don't think there would be enough value to do it myself.

Patrick