[Cogl] Cogl in Python

[Artur Skawina]

On 04/11/12 02:10, Martin Owens wrote:
> 
> 
> On Wed, 2012-04-11 at 02:01 +0200, Artur Skawina wrote:
>>
>> so the clutter and cogl GI parts seem ok, at least here. 
> 
> Yes, we know this works from the C because it's the same instruction
> that happens in the void clutter_texture_paint(); And isn't that using
> gobject?
> 
> What's a Q&D?

Quick and dirty, as i didn't really create a new class. I'm just pointing
out that the clutter and cogl GI data is enough to build working language
bindings. It's not perfect, requires some fixups, does not mark all
constructors as such, uses Handles instead of the proper types etc.
But your problem seems to be inside the python bindings, which i'm not
familiar with. ISTR that the cogl module was the hardest one to turn from GI
to working D module - so I initially thought that maybe there was some error
in the GI data, that i didn't notice because i never used these parts of
the API. Things like missing '*' in types or parameters happen. But as the
same APIs work from D, which uses *only* the GI data to call the C functions,
the bug seems to be elsewhere. 
Hmm, get_cogl_material() returns an Handle, which is a (void*), so if you're
seeing segfaults when calling every Material method maybe the python bindings
somehow fail to convert that.

artur