Re: GStreamer regression analysis [was: GNOME and GStreamer]

[Bastien Nocera <hadess hadess net>]

On Mon, 2006-01-16 at 21:20 +0100, Thomas Vander Stichele wrote:
> 
> Of course, that's the point I agree with.  I'm just saying that this
> is
> not a fair reason to call all of GStreamer 0.10 a development series.
> The core is stable, as are the base plugins.  Additions can be made,
> as
> can new plugins.  How is this different from GTK+ 2 ? How long did it
> take to get an even halfway decent file selector in ? The only
> difference is that us still being a 0.whatever is precisely because we
> haven't commited to a 1.whatever API stability yet. 

Har har. The file selector.
Using the new file selector was a choice that application developers
made, and the old fileselector was not a regression, it was the same
crappy one as the older releases.

The main difference is that GTK+ 2.x was evolutionary, and GStreamer's
move from 0.8.x to 0.10.x isn't. Maybe it's needed, I'm not to judge.
The problem is the regressions in playback for most filetypes (I'm not
talking about the game formats), and non-working parts of Totem like the
properties, the thumbnailer and the Mozilla plugin.

I wouldn't recommend anyone ship with the GST 0.10 backend by default if
most of those regressions aren't fixed. I really don't care for 1.0 API
stability, but I care for 1.0 stability and reliability.

---
Bastien Nocera <hadess hadess net>