Re: Hard freeze exception for two regressions from pygobject 3.1.92

[Matthias Clasen <matthias clasen gmail com>]

On Wed, Mar 21, 2012 at 12:39 PM, Martin Pitt <martinpitt gnome org> wrote:
> Hello release team,
>
> (CCing the other active maintainers of pygobject).
>
> We released pygobject 3.1.92 on Monday, and it turned out to have two
> regressions:
>
>  * Marshalling of GHashTables that contain GValues inadvertedly got
>   broken by the fix that makes pygobject work on big-endian machines:
>
>     https://bugzilla.gnome.org/show_bug.cgi?id=668903
>
>   This is fixed with
>
>     http://git.gnome.org/browse/pygobject/commit/?id=efcb4b0b3
>
>   which is a straightforward patch (also attached for your
>   convenience). If you prefer, I can also reduce the cherry-pick to
>   just adding the missing "case GI_TYPE_TAG_INTERFACE:", and drop the
>   g_assert change for the stable branch.
>
>   We also have a new test case for this which will ensure that it
>   won't break again. These are attached to above bug report, but due
>   to the current hard freeze we can't push them into
>   gobject-introspection git yet, so they will need to land after the
>   freeze.
>
>   I strongly recommend to allow this, as it is a significant regression.
>
>
>  * When using the Gtk.TreeStore.append() API (same for ListStore, and
>   the prepend() and related methods), and you supply None values for
>   columns, you now get warnings. They don't actually break your
>   program, but nevertheless look ugly:
>
>     https://bugzilla.gnome.org/show_bug.cgi?id=672463
>
>   This was introduced with
>
>     http://git.gnome.org/browse/pygobject/commit/?id=bf8c95836e1c
>
>   which changed the behaviour from setting non-None columns to
>   setting them all. The fix already landed in master:
>
>    http://git.gnome.org/browse/pygobject/commit/?id=8d85d66
>
>   and I would like to apply it to the pygobject-3-2 branch (which is
>   what GNOME 3.4 will release with) as well. For convenience I attach
>   the patch as well.
>
>   This code is well-covered with tests:
>
>     http://git.gnome.org/browse/pygobject/tree/tests/test_overrides.py?id=8d85d66#n852
>
>   onwards, so I am very confident that this does not break anything.
>
>   If you are hesitant about this one, I'm also ok with not
>   cherry-picking this, as it's mostly a cosmetical issue (but might
>   flood .xsession-errors and debug logs with bogus warnings).
>
> Thank you in advance!

Looks ok to me; if the bindings guys are in favor, +1 for the release team