merging PyGI into PyGObject

[Tomeu Vizoso <tomeu vizoso collabora co uk>]

Hi,

Johan has suggested merging PyGI back into PyGObject because:

- PyGI has reached "useful" status, applications that make use of it are 
packaged and used in distros and we are close enough to being 
feature-complete.

- PyGI is more mature now: is regularly maintained, uses the GNOME 
infrastructure, is packageable, code contributions are reviewed, has 
consistent code guidelines, etc.

- some of the biggest features missing in PyGI overlap with 
functionality already in PyGObject, namely properties and signals. If we 
want to be able to support properties that return collections and do so 
with proper memory management, we need to be able to use the information 
available through introspection. If we tried to implement this in PyGI, 
we would be duplicating lots of work currently in PyGObject.

- some changes are needed in the GType system for supporting 
fundamentals and foreign structs coming from g-i.

It has been proposed as well making the --enable-pygi switch the default.

Opinions?

Regards,

Tomeu