Re: [Banshee-List] About GstPlayerEngine

[Julien Sobrier <julien sobrier net>]

Aaron Bockover wrote:
> All the "capabilities" are essentially hints. I'll have to double check
> about mms, but basically it works like this for both codec and transport
> support:
> 
> If a player engine _explicitly_ declares support for something, it has
> preference. Otherwise, the default engine is used. For instance, MP3
> preference is given to the Helix Remote engine (RealPlayer). If Helix
> Remote is not installed, the default (GStreamer) engine will be used.
> Just because something isn't in the list doesn't mean it's not
> supported.
> 
> Regarding gstreamer-sharp: Khaled is starting to pick up more work on
> the project and we'll make information more public in the near future.
> He is working on a few separate goals which will eventually tie together
> in the bindings. After extensive unit tests and QA, yes, the plan is to
> drop libbanshee altogether and port all the C backend GStreamer code to
> C#.
> 
> The goals are basically this:
> 
> a) Use GAPI to generate bindings, fix up API and add customs/glue where
> necessary (mostly done)
> 
> b) Port PyGst unit tests to C#, write more unit tests where
> necessary/specific to mono bindings (partially done)
> 
> c) Implement dynamic signal support necessary for binding managed
> delegates to arbitrary GLib/GObject signals at runtime (essentially
> done, need to support GLib.Opaque/GstMiniObject marshalling)
> 
> d) Port official GStreamer samples and possibly a backend or two from
> libbanshee as tests (very little done)
> 
> Cheers,
> Aaron

Thanks a lot.

It would be great to see the raw code generated by GAPI and the final
code to see what had to be changed manually. I think it would be very
helpful for people who wants to create bindings for other C libraries.

Talking about unit testing, is there any of them for Banshee. I'm
wondering how useful and feasible it would be to add unit tests to an
application such as banshee. I think it is easy to test for unhandled
exceptions, but how can we verify that there is actual sound produced,
or that the UI reflects the right things, etc.?

BTW, MMS works with "Open location", except that the buffering is not
done well.

Julien