On 05/12/2016 08:27 PM, Robert McQueen wrote: > I don't understand how such a mismatch would come about "accidentally": > if you are maintaining/building your app against the 3.20 runtime and > you know GtkSourceView 3.20 is therefore needed, you should select that > version to bundle. > > The flatpak you've built will never (automatically) be run against 3.22 > unless/until you do some maintainer action to say, here is my 3.22 > release, I want the 3.22 runtime and so therefore will bundle > GtkSourceView 3.22. > > So ensuring that something in a bundle matches something in a runtime is > not by itself a reason to put something into the runtime - what to > bundle and what runtime to depend on is an explicit choice of the bundle > maintainer (and in GNOME we very much hope that bundle descriptions are > either maintained by the module owner, or generated by the release team > a la jhbuild, or some combination thereof).
I know I'm certainly not the only person who ships buggy .0 software. I assume we do plan on updating with dot releases of GNOME? Should we work on libabigail automated testing? Obviously there are things that libabigail can't check (like g_signal_new_class_handler) and properties. Sounds like useful tooling for continuous, even. -- Christian _______________________________________________ gnome-os-list mailing list [email protected] https://mail.gnome.org/mailman/listinfo/gnome-os-list
