[vlc-devel] Re: vlc: svn commit r20223 (courmisch)

[Rémi Denis-Courmont]

Le lundi 21 mai 2007, Damien Fouilleul a écrit :
> Subversion daemon wrote:
> > r20223 | courmisch | 2007-05-19 23:43:16 +0200 (Sat, 19 May 2007) |
> > 2 lines Changed paths:
> >    M /trunk/include/vlc_common.h
> >    M /trunk/modules/audio_filter/converter/mpgatofixed32.c
> >    M /trunk/modules/audio_output/alsa.c
> >    M /trunk/modules/codec/faad.c
> >    M /trunk/modules/codec/ffmpeg/encoder.c
> >    M /trunk/modules/codec/ffmpeg/ffmpeg.c
> >    M /trunk/modules/codec/ffmpeg/postprocess.c
> >    M /trunk/modules/codec/ffmpeg/scale.c
> >    M /trunk/modules/codec/libmpeg2.c
> >    M /trunk/modules/codec/x264.c
> >    M /trunk/modules/codec/xvmc/xxmc.c
> >    M /trunk/modules/stream_out/switcher.c
> >    M /trunk/modules/video_filter/deinterlace.c
> >    M /trunk/src/libvlc-common.c
> >    M /trunk/src/misc/cpu.c
> >
> > Use a vlc_CPU() wrapper instead of (ab)using libvlc_global
> >
> >  > http://trac.videolan.org/vlc/changeset/20223
>
> although i do appreciate this commit, i think vlc_CPU() context
> should be per instance, rather than global, because a user can turn
> off MMX, or FPU capabilities on a per libvlc instance basis

I am very unconvinced whether there is any serious use-case for that: 
some weirdos might DISABLE(!) optimizations, but why on earth for one 
of multiple instances within a single process?? That being, so long as 
you do not expect me to do it, I don't mind :P Note that CPU features 
have always been per-process so far.

Basically, vlc_CPU() would need to take a vlc_object_t *, and flags 
would be stored in libvlc_int.

-- 
Rémi Denis-Courmont
http://www.remlab.net/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://mailman.videolan.org/pipermail/vlc-devel/attachments/20070521/4e7f14df/attachment.sig>