Flatpak support in CI

[Ben Cooksley]

Hi all,

This evening I have fixed the broken Flatpak builds.

This breakage was caused by a regression shipped by Flatpak itself, which
Fedora incorporated and was too slow in shipping the fix for. See
https://bugzilla.redhat.com/show_bug.cgi?id=2528386 and
https://github.com/flatpak/flatpak/issues/6818

It only surfaced for us recently because we had previously been using an
older version of Fedora for our build image.

Updating that was necessary to get a new version of flatpak-builder, which
fixed yet another regression that was causing the Flatpak KDE Runtime to
fail to build.

We had up until the Fedora update been forcibly downgrading from the most
recent Flatpak Builder version available (which was old) to get a version
that did not have the breakage in it to keep that working.

This is a situation that is unsustainable, and the shipping of two
regression containing releases in the last few months is, quite frankly,
unacceptable. Not to mention a significant waste of my time when people ask
me why it is broken - for which there are much better uses.

Those who are advocates for Flatpak should have a serious discussion with
the Flatpak developers regarding their release QA processes, which are not
sufficient to catch regressions with significant impact.

I would suggest they incorporate builds of some KDE Flatpaks as a minimum
baseline.

Should this pattern of Flatpak releases containing severe regressions
continue we will need to revisit the subject of Flatpak CI with a view to
terminating it, including the Runtime builds.

Regards,
Ben
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/release-team/attachments/20261007/60c2c837/attachment.htm>