[gst-devel] Results from some quick tests of Totem

[Martin Soto]

On Tue, 2004-03-23 at 17:35, Ronald Bultje wrote:
> Why waste developer time on a dead thing?

Agreed. Opt is in a messy state right now, and I don't think it'll be
easy/worthwhile to try and take it out of it.

> If people cared, they'd write a
> new one, document the old one or anything but the current situation.

Ask and you will be given. I have a new scheduler that I wrote last
week. I call it the fair scheduler, due to the algorithm it uses. It can
play mp3 files for me, and it even runs my DVD playing pipeline. It
still has some issues, though. I implemented nonblocking clock waiting
and it works, but it seems to cause some roughness every now and then
(or at least, is the only reason I can see for the roughness). I need to
debug that, anyway. And sometimes it spits funny GObject messages when
applications are shut down, which is probably due to issues in the Omega
threads implementation, but I'm not yet sure.

Anyway, I'll put a patch on the list soon, probably tomorrow. I'll be
glad if you test it and report any problems back to me.

> I know we cannot switch, but that doesn't make opt any less broken. I have
> yet to encounter my first bug with basicgthread in gst-rec, totem or RB.

I'd expect basic to work better, because it uses a simpler algorithm. My
fair algorithm is even simpler, implementing it brought me continuous
memories of my OS class many years ago...

Enjoy,

M. S.
-- 
Martin Soto <soto at informatik.uni-kl.de>
Universität Kaiserslautern