From: Bastien N. <ha...@ha...> - 2002-09-14 15:08:52
|
On Sat, 2002-09-14 at 10:27, Andres Salomon wrote: > On Sat, Sep 14, 2002 at 03:44:56PM +0200, Guenter Bartsch wrote: > >=20 > > hi andres, > >=20 > > On 09/14, Andres Salomon wrote: > >=20 > > > I've been meaning to ask this; is there any reason xine-utils is used > > > (and wheels reinvented), instead of using something like glib? Only > > > reason I could see was due to processor-specific optimizations. > >=20 > > i wasn't aware that glib offers a significant part of the functionality > > xine-utils does? xine-utils mainly contains mmx-macros and other lowlev= el > > assembly stuff and besides that a few utility functions (oki, a linked > > list and maybe xmalloc can be found in glib). >=20 > The list, malloc, various utility functions (everything in utils.c), > xml parsing stuff, and thread stuff are all implemented in glib. That > leaves the per-arch memcpy optimizations, and the YUV stuff. XML parsing could use libxml or g_markup depending on the level of complexity of the xml data to decode. > > besides that the goal is of course to keep external dependancies to a > > minimum >=20 > I think glib is common enough (on unix systems, anyways) that it > wouldn't be a major problem to depend upon. However, I know xine folks > have spoke about embedded system stuff; I'm not sure how well glib would > fare there. Although this would be far from perfect, there's always the cut'n'paste solution. We already do it for a number of libraries. Using bits of glib would be easy enough I guess. That's going to be more complicated if some bits require GObject though. Cheers --=20 /Bastien Nocera http://hadess.net |