[FFmpeg-cvslog] r22052 - in trunk/libavcodec: h264.c h264.h

Michael Niedermayer michaelni
Thu Feb 25 23:40:05 CET 2010


On Thu, Feb 25, 2010 at 10:19:20PM +0000, M?ns Rullg?rd wrote:
> Michael Niedermayer <michaelni at gmx.at> writes:
> 
> > On Thu, Feb 25, 2010 at 10:49:18PM +0100, Michael Niedermayer wrote:
> >>     -On Thu, Feb 25, 2010 at 07:55:36PM +0000, M?ns Rullg?rd wrote:
> >> > Michael Niedermayer <michaelni at gmx.at> writes:
> >> > 
> >> > > On Thu, Feb 25, 2010 at 07:20:18PM +0000, M?ns Rullg?rd wrote:
> >> > >> Michael Niedermayer <michaelni at gmx.at> writes:
> >> > >> 
> >> > >> > sorry again, ill try not to forget svq3 from now on when compiling with
> >> > >> > disabled dependancies. (enabling deps is no option as its too slow)
> >> > >> 
> >> > >> Can you please elaborate on that?  If you disable deps, you _will_
> >> > >> miss things like this.  I'm not aware of any case where more than
> >> > >> necessary gets recompiled.
> >> > >
> >> > > add a comment to avcodec.h, bump minor api version, ...
> >> > > there are many things that cause quite a bit to be recompiled
> >> > 
> >> > It's hard to track dependencies other than by file, so yes, that would
> >> > trigger some superfluous rebuilds.  However, this change was only to
> >> > h264.h.  Nothing irrelevant depends on it.
> >> > 
> >> > > was there some way to quickly enable/disable deps?
> >> > > currently i have common.mak hacked to achive it, but i faintly remember
> >> > > some discussions about it ...
> >> > 
> >> > make DEPS=n
> >> 
> >> Thanks ill try to build with deps from now on before or immedeatly after
> >> a commit so such hour long breakage does not reoccur
> >> 
> >> 
> >> > 
> >> > Before anyone asks, this is a hack that should not be documented.  I
> >> > don't want to deal with the trouble reports that would ensue.
> >> > 
> >> > > also is there some way by which make would display what it would rebuild
> >> > > without doing a time consuming rebuild? (had i run this with deps i would
> >> > > have spot svq3)
> >> > 
> >> > make -n
> >> 
> >> not readable at all without filtering it though sed or other
> >> btw here it take 7-11 seconds, which isnt exactly fast.
> >
> > and thats without deps 
> > with deps its 21 seconds
> >
> > and while my pentium dual does it in 2 secs with deps and 1.3 without i
> > still think this is ridiculously slow for what it does
> 
> On my computers it takes but a fraction of a second.  What filesystem
> are you using?  

ext2/ext3
suggestions for something faster are welcome


> Is it any quicker with make -r?

time make -n -r >/dev/null

real	0m1.722s
user	0m1.453s
sys	0m0.159s

time make -n >/dev/null

real	0m12.078s
user	0m11.177s
sys	0m0.252s

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Everything should be made as simple as possible, but not simpler.
-- Albert Einstein
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-cvslog/attachments/20100225/8a6bf81b/attachment-0001.pgp>



More information about the ffmpeg-cvslog mailing list