Re: Target state of gnome-build-meta

Hi,

On Wed, 2019-05-01 at 14:58 +0200, Karl-Philipp Richter via gnome-devel-list wrote:
> Hi,
[...]

> I'm investigating unused potential which I see in the current processes
> and seeking to prepare an overview as basis for a discussion. In order
> to avoid unnecessary work I'm looking for an eventual definite statement
> about the target state of
> > [gnome-build-meta](https://gitlab.gnome.org/krichter/gnome-build-meta)
> in terms of stability of the master branch. If there's one defined, are
> build failure allowed in this target state (no matter how it looks now)?
> Are merge requests allowed to be merged before CI passed?

It is rather a work in progress, we use gnome-build-meta for the
following things:

 - Making releases

   We build core.bst to ensure that everything builds with every
   release, and we convert the git repos in the repository into
   tarballs and publish a project which will produce exactly the
   same results as when we built it at release time.

 - Building the GNOME flatpak SDK


Currently I am working on making it build a bootable image of latest
GNOME too.

About gating commits to gnome-build-meta on CI, yes we do that, but no
it is not reliable - we need to bring some more clarity to this process
indeed.

The thing is, we run `bst track` as a part of the CI, which means that
we build "the latest master of every GNOME module" every time, so a
commit to the gnome-build-meta module (defining how things get built)
is not guaranteed to be the only thing which can cause CI to break - a
change in some of the modules can also break CI.

Really, there are better ways to do this, but I'd rather not get into
it on this list as I suspect the interested parties are not on this
list (I really don't see a lot of traffic here).

Maybe it would be beneficial if you were to come and ask about this in
the #freedesktop-sdk channel on freenode, or in the #release-team
channel on GNOME IRC ?

Cheers,
    -Tristan