From: M G. <el...@al...> - 2002-02-13 00:31:34
|
> Just my 2 cents about the pthread issue. I don't know how much of pthread is > actually used in xine, but if you don't care about 100% correctness and 100% > implementation of the pthreads specs, then implements the bits you are > interested in. > We've done this for videolan, an apart from being quicker (we didn't care > about correctness) it doesn't represent much code either. > This was actually my first thought when I started the xine port. The plan was to write an abstracted thread api. This is what I have done in the past with my own projects that compile on both unix and win32. However, the xines buffer mechanisms rely heavily on pthread conditional locks and some threads also uses attributes from time to time. This basicly threw me back to reimplimenting most of pthreads on win32. I would much prefer an abstracted api ( and maybe someday I will put in the time to write one if everyone else agreed it was a good idea ). I definately think abstraction would be the best solution for a time sensative media app. But for now the pthread win32 port is the way to go as far as I am concerned. btw, Did you say you worked on the vlc code? I enjoyed reviewing your projects source very much. I found it very comprehensive and well documented. Cudos to the videolan team for such a great product. -Matthew Find the best deals on the web at AltaVista Shopping! http://www.shopping.altavista.com |