[Ffmpeg-devel] [PATCH 8/9] Doxygenize av_fast_realloc comments

[Michael Niedermayer]

Hi

On Thu, Feb 22, 2007 at 10:32:31PM +0100, Panagiotis Issaris wrote:
> Hi Michael,
> 
> Panagiotis Issaris schreef:
> >Michael Niedermayer schreef:
> >>On Thu, Feb 22, 2007 at 08:47:58PM +0100, Panagiotis Issaris wrote:
> [...]
> >>>>additionally, this function is a internal function of avcodec
> >>>>and should not be used outside, it shouldnt have a av_ prefix  but ff_
> >>>>and it should be droped if normal av_realloc() is as fast (benchmark a
> >>>>large number of reallocs)
> >>>> 
> >>>>      
> >>>I?ll first prepare a patch for moving it and doing the s/av/ff/ 
> >>>thing. Afterwards I'll post
> >>>some benchmarking code so that others correct it and run the 
> >>>benchmark too.
> >>>    
> >>
> >>hmm wouldnt it be better to benchmark it first before doing a possibly
> >>redundant move/rename patch?
> >>  
> >Yes, but I thought it might take a while till consensus is reached on 
> >whether it is faster or not, and
> >that it might also take awhile before it gets benchmarked on enough 
> >different architectures/CPUs.
> A first attempt of some benchmarking code for av_fast_realloc() 
> hideously embedded within FFmpeg's main function.
> 
> I do not know if I'm benchmarking the correct type of usage of 
> av_fast_realloc() vs av_realloc() nor if I am using
> START_TIMER|STOP_TIMER in the way you had intended it...
> 
> For each test of av_fast_realloc() I tested two runs, one with 
> size<min_size and one with size>min_size, to test both
> cases. (Besides the obvious av_realloc() run ofcourse.)
> 
> _Obviously_ not intended for Subversion.

you change the internal size variable for av_fast_realloc() so the result
is meaningless at least as far as i can see, the size variable is never
ever chnanged by anything except av_fast_realloc() itself

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

I do not agree with what you have to say, but I'll defend to the death your
right to say it. -- Voltaire
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20070223/e2a9ac55/attachment.pgp>