[Rhythmbox-devel] the future for Rhythmbox

[Benjamin Otte <in7y118 public uni-hamburg de>]

Hi people,

The following text is an answer I posted to gnomedesktop.org when someone
asked about how new maintainers affect upcoming releases.

Start copy/paste here:
======================
If it affects a release at all, it will only affect it positively, because
if someone is the maintainer who doesn't want to, it's not going faster.
;)

That said, there is no clear direction for Rhythmbox right now that I know
of. I was even quite a bit surprised that I get announced as
co-maintainer. I had talked to Colin about being interested and it seems
it somehow ended up in the summary. (Not that I'm opposed to it, mind
you).
I think in the short term (measured in Rhythmbox time units :/) Colin will
want to get
a new release out, so the Rb/net-Rb confusion ends and people can use the
new UI.

But we'll all have to think about the long term.
Let me give you an overview from my perspective.
Things that are great about Rhythmbox:
- UI.
The current Rhythmbox cvs interface absolutely rocks the boat. And there
are still ideas on the mailing lists to make it even better.
- GStreamer
It's great the RB doesn't try to reinvent the wheel and uses the GNOME
multimedia framework.
Things that are not so great about Rhythmbox:
- The code
The Rhythmbox code suffers from feature bloat and missing organization a
lot. This probably comes from when people were still not sure about what
Rhythmbox really is supposed to do and how it's supposed to look and they
hacked features into the code. There is no good GUI/backend seperation for
example.
This is bad for new developers that want to get going with Rb. And it is
_really_ bad because the only comments in the code are the license
disclaimers at the top.
- Monkey-Media
It's basically a thin wrapper around GStreamer that hacks around problems
/ missing features. While doing so it introduces a lot of problems on its
own. (Think of it as drop shadows for GStreamer)
- GStreamer
Rhythmbox was the first bigger application to ever use GStreamer. There's
still only one and a half other apps for GStreamer (a-j and totem). If
you're the first to try something you find the problems. Rhythmbox found a
lot of problems in GStreamer and still finds new ones.

What I would do with Rhythmbox right now if I had total control would be
the following in that order:
1) Make Colin do a release of current RB.
2) get missing features into GStreamer HEAD. Top problem here is tag (aka
metadata) support.
3) Make GStreamer development release(s) of HEAD.
4) Throw out broken features to reduce code size (Bonobo, CD support, ...)
5) Ditch monkey-media
6) Start replacing the backend based on HEAD GStreamer and GNOME/Gtk 2.4.
This time with comments and an idea of how the code should look.

This requires some things though: Me really being the co-maintainer (I'm
still confused by this), RB devs who agree with all that, developers who
implement that and - at least on my part time - which I will have very
little the next weeks because of work.


Woo, this was a long post now, I think I'll send that to RB-devel tonight.

Benjamin

===================
end copy/paste here


So before anyone has problems with it: I don't consider myself to be a
maintainer of Rhythmbox yet, but I'd sure like to co-maintain it with
Colin.
Are you all fine with that?

And what do you think about them ideas I expressed in that post?


Benjamin