Re: [Rhythmbox-devel] Rhythmbox archive

[Jonathan Matthew <jonathan kaolin hn org>]

On Tue, Jun 07, 2005 at 11:40:43PM +0100, Bastien Nocera wrote:
> More updates.
>
> [...]
>
> > patch-13
> >     HACK: work around broken shoutcast servers not responding to HTTP
> > HEAD requests
> 
> That obviously won't make it. Is there a bug number/test case for this?
> The same problem was happening with Totem, and it got fixed with a more
> recent gnome-vfs, and the neon HTTP backend.

Test case is pretty simple:  try to add
'http://server2.somafm.com:8000/listen.pls' as an iradio station.  It
appears, sits there for a while, then disappears.

That isn't the URL that www.somafm.com tells you to use, but there are a
few streams around that do use the '/listen.pls' from the shoutcast/icecast
server.

I'm using gnome-vfs 2.10.1 (Debian package from experimental) with the
neon HTTP backend.

> > patch-15
> >     drag/drop of iradio streams from browsers into radio source
> > patch-16
> >     set iradio stream name and genre from stream metadata if not
> > already set
> 
> Doesn't that clash with the port to playbin?

Sort of.  It doesn't work with playbin at the moment, but it (or a
variation of it) should eventually.

> > patch-17
> >     make setting iradio stream properties from stream metadata
> > actually work
> 
> Is that on top of patch-16?

Yes.

> > patch-23
> >     fix amg urls (perhaps), use urls from artist/album fields
> > (netlabels)
> 
> Committed (I fixed the non-C89 declaration in there :)

Some day I'll learn to actually check for these things..

> > patch-24
> >     make file choosers not local-only again (again?)
> 
> Committed the API change part, but I set the defaults to be non-local,
> as it deadlocks with the good ol' "sftp with password" trick.
> Totem test case here (we used to hang as well, but I removed all use of
> gdk_enter and leave_threads):
> http://bugzilla.gnome.org/show_bug.cgi?id=154796
> Original bug:
> http://bugzilla.gnome.org/show_bug.cgi?id=162989

Fair enough.

> > patch-39
> >     possibly fix weirdness with cursor position jumping while editing
> > search box (#171008)
> 
> As I can't reproduce the problem, I won't commit this one just yet (only
> appears on "slower" machines?)

I can reproduce it on a 3GHz P4 with around 10000 songs in rhythmdb.
It's entirely subjective, but I find the search box is much easier to
use with this patch in place.

> > patch-44
> >     apply playlist manager dirty flag fix (necessary for something
> > else i'm about to fix)
> 
> What is this last one needed for? Is it OK to merge even though it's
> only part of a larger patch? (And what would this larger patch do?)

It was needed for patch-45 (which doesn't seem to be in Christophe's
patch pile), which fixed #304519.  It's useful on its own, as it fixes
#166340.

> Pennies to the person who manages to track down:
> http://bugzilla.gnome.org/show_bug.cgi?id=306819

Is this the one where you get some strange ORBit-related messages on
stdout?  Right on cue, it just happened again:

** ERROR **: file poa.c: line 1025 (ORBit_POA_activate_object_T): assertion failed: ((poa->life_flags & ORBit_LifeF_DeactivateDo) == 0)
aborting...

This one has always scared me too much to seriously investigate it.

thanks,
-jonathan.