Re: Master moves to Qt 6 this Wednesday

Ingo Klöcker <[email protected]> Tue, 12 Sep 2023 09:57:58 +0200

Newsgroups	gmane.comp.kde.devel.kwrite
Message-ID	<2904085.e9J7NaK4W3@daneel>

On Dienstag, 12. September 2023 00:44:20 CEST Albert Astals Cid wrote:
> El dilluns, 11 de setembre de 2023, a les 23:44:01 (CEST),
> [email protected] va escriure:
> > On 2023-09-11 21:42, Albert Astals Cid wrote:
> > > Nothing else is broken but kate (this is my past 3 weeks status report
> > > speaking, current point in time may be different), so no you're not
> > > responsible to fix all global CI breakeages, you're responsible of
> > > fixing kate
> > > CI.
> > 
> > that is not very fair and if the CI is disabled build failures will
> > creep in.
> 
> CI job is still there so no build failure will creep in, it's only that
> Windows/Qt6 failing tests are not considered a fail anymore, so basically
> the same status as before but now it's marked as passing instead of
> failing.

Maybe you notice failing tests in green jobs. I certainly don't. For me 
passing CI means everything is alright including all automatic tests.

On the other hand, I do agree that unstable tests are highly annoying 
especially for "external" contributors who don't know about them and try to 
figure out if they did something wrong in their MR.

And then there are tests that start failing for no apparent reasons in the 
project because they are caused by external changes, e.g. upstream ABI 
breakages. A project maintainer has little to no influence on this and I agree 
with Christoph that it's unfair to blame the project and its maintainers for 
this. There's no good solution for this, but failing CI hopefully at least 
keeps those cases on the maintainers' radars.

> Once the tests are passing again the change can be put back to consider
> Windows/Qt6 failing tests a failure of the job.

I have prepared a draft MR for this, so that we can more easily see when the 
tests are passing again:
https://invent.kde.org/utilities/kate/-/merge_requests/1298

Regards,
Ingo

signature.asc (application/pgp-signature, 228 B)