Re: GIR-based vala bindings

[Jiří Techet <techet gmail com>]

On Fri, Apr 8, 2011 at 23:49, Lorenzo Masini <lorenxo86 gmail com> wrote:
> Il giorno gio, 07/04/2011 alle 23.42 +0200, Jiří Techet ha scritto:
>> What does it mean that the vapi's weren't good? How did they look like
>> (e.g. when diffed against those distributed with libchamplain)? Was
>> something missing inside? Did you get any errors or warnings during
>> their generation?
>
> Diffs attached.

Hm, no idea what's causing the difference. How about the girs, are
they identical in both cases? (i.e. isn't this problem rather related
to some changes in the introspection scanner?)

>
>
>> I don't plan to make a release just with this single change - I don't
>> find it that critical (vapi generation works fine in my opinion and in
>> the worst case the package can be created by manually copying the vapi
>> files from the tarball) so I'll wait for some time for other bugs to
>> appear and make the release later. It would be better to find out
>> what's causing that the vapi files aren't generated correctly on your
>> system.
>
> Uhm, another change could be to fix the macro to get the type of the
> GtkChamplainEmbed to follow the nomenclature NAMESPACE_TYPE_CLASS. Now
> it's GTK_TYPE_CHAMPLAIN_EMBED. The macro for GtkClutterEmbed type is
> GTK_CLUTTER_EMBED, so the one for champlain-gtk should be
> GTK_CHAMPLAIN_TYPE_EMBED.

Oops, I've never noticed it, that's definitely wrong - I've fixed it
in master but this changes the API so I'll keep it as it is in the
0.10 branch. But does this change fix your problem? For instance the
demo we have uses only libchampplain without gtk so this change
shouldn't have any effect.

Cheers,

Jiri