Re: Completed what I can in gstreamermm

[José Alburquerque <jaalburquerque cox net>]

Murray Cumming wrote:
> On Wed, 2008-05-07 at 14:09 -0400, José Alburquerque wrote:
>   
> > > > > I wanted to let everyone know so that others are aware that there is 
> > > > > still more to be done and the project can continue.  I don't think I'll 
> > > > > be needed because all wrapping should be really clear, but if there 
> > > > > should be any questions please don't hesitate to e-mail me.  Thank you 
> > > > > all for the opportunity to exercise my programming muscles.  Take care.
> > > > >     
> > > > >       
> > > > >           
> > > > While preparing the NEWS file for gstreamermm-0.9.6, I was amazed at how
> > > > much work you've done, some of it really difficult stuff.
> > > >
> > > > The media_player_gtkmm example shows how much you have achieved.
> > > > Actually playing music and video is the main point of gstreamer.
> > > >         
> > I didn't mention:  I plan to use some of the media_player_gtkmm code in 
> > a project I'm developing.  Now that I was able to tackle some 
> > challenges, it will come in handy when I work on the other project. :-)
> >     
>
> I done some cleaning up of the API. It's mostly just small changes which
> won't require much changes in your application code. For instance, it's
> now ElementFactory::create_element() instead of
> ElementFactory::create().
>
>   

An obviously useful logical change. :-)  Changes like these will not 
affect my application development because I understand that gstreamermm 
is still in early development so I'm happy to adjust my project 
development accordingly while gstreamermm stabilizes.

> However, I am attaching the patch for a change I made to Structure, to
> remove the awkward taking-ownership stuff. This is theoretically now
> doing more copying than really necessary (it's a pity that it's not
> reference-counted), but it makes the API much nicer. I wonder if this
> causes problems for you.
>
>   

To be honest, I haven't begun to use gstreamermm enough yet in my 
project for this change to affect me yet, but looking at it more 
closely, there was an error in my logic.  I was trying to ensure that if 
someone tried to modify a Gst::Structure of a Gst::Caps obtained with 
Gst::Caps::get_structure() the change would be reflected in the actual 
Gst::Structure of the Gst::Caps as in the following:

...
Gst::Structure structure = caps->get_structure(0);

if (structure)
{
 Glib::Value<int> width;
 width.init(Glib::Value<int>::value_type());
 width.set(300);

 structure.set_field("width", width);
}
...

I had unfortunately overlooked that the GstCaps docs say that 
GstStructures of GstCaps should not be modified if obtained with 
gst_caps_get_structure(const GstCaps*, guint index) 
<http://gstreamer.freedesktop.org/data/doc/gstreamer/head/gstreamer/html/gstreamer-GstCaps.html#gst-caps-get-structure>.  
What you're doing is best and if I had been more careful reading the 
docs, I would not have made these changes to Gst::Structure.

> I might need to do something similar for TagList.
>  
>   

As you say, TagList should undergo the same changes.

BTW, I had taken a look at your commits of last week and noticed some 
TODO's that I might have quickly been able to do (time notwithstanding) 
but I could not because I was affected by the svn+ssh bug.  I'm still a 
bit pressed for time, but if I find time here and there, would it be 
alright if I commit some light stuff also?

--
José Alburquerque
jaalburquerque cox net

The path to real wisdom begins with a deep loving respect for the ever living God (Prov. 1:7)