[FFmpeg-devel] [PATCH] configure: enable libavresample by default

[Michael Niedermayer]

On Mon, Oct 21, 2013 at 11:10:42PM +0200, wm4 wrote:
> On Mon, 21 Oct 2013 22:15:12 +0200
> Michael Niedermayer <michaelni at gmx.at> wrote:
> 
> > On Mon, Oct 21, 2013 at 08:53:24PM +0200, wm4 wrote:
> > > On Mon, 21 Oct 2013 20:34:49 +0200
> > > Michael Niedermayer <michaelni at gmx.at> wrote:
> > > 
> > > > Thats a long list, but i think for discussing the change of
> > > > avresamples default build status, theres one keypoint missing
> > > > and that is who would maintain it ? 
> > > 
> > > Libav. You're not making any changes to libavresample, so merging
> > > Libav's maintenance patches is no additional work.
> > 
> > So you do not volunteer to do the work and you explain me how it will
> > be no additional work for me? (why dont you volunteer if its no work?)
> > and
> > you seem to assume we dont fix any security issues, fix no bugs,
> > review no patches and hope it will still build with our build system
> > in the future.
> 
> But just disabling libavresample makes all this better?

> 
> What if someone enables libavresample using --enable-avresample, and
> there's a bug?

> Will you fix it?

i cant say it for absolutely certain as i dont know the exact
situation, but most likely not


> Will you just say you don't care
> because it's not enabled by default?

How do i get the honor to be connected to this ?
If someone posts a bug report to trak about avresample, every developer
ffmpeg and libav developers alike can see it, can reply, can work on
and fix it.
If noone does, well then noone does.


> 
> > In reality, libavresample doest even pass its selftests in ffmpeg
> > thats today, no future hypothethical stuff
> > who will fix that ?
> 
> You're saying that libavresample doesn't pass the tests Libav added?
> Which I find funny, because your commit adding the libswresample tests
> say they're based on libavresample's - so you spent some work on it,
> but the benefit went only to libswresample, not libavresample.

so from "is no additional work" we now reached a mild accusation of
why iam not doing additional work 

yes i looked at the avr fate tests and tried to enable them
many/most/all? failed, i then changed the implicit aresample use to
explicit resample (which would end up actually using avr) and tried
again and it failed again.
At that point and after seeing how much twiddling libav needed to
get it to pass on all architectures with increasing fuzz factors
i decided its not worth it and just worked on the next item i
wanted to work on which is swr fate tests



> 
> > 
> > I do maintain swresample, if it doesnt work i will fix it, if
> > avresample is better in some way i will improve swresample.
> 
> Then give it exactly the same API as libavresample.

writing a wraper around swr that is compatible with the avr API/ABI
is something that can be done, iam happy to help and iam happy to
apply patches in that direction. Iam not a user of that though so
iam probably not going to do it alone

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Asymptotically faster algorithms should always be preferred if you have
asymptotical amounts of data
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20131022/20d1b1e5/attachment.asc>