[FFmpeg-devel] [PATCH] avutil/utils: remove unnecessary things

Michael Niedermayer michaelni at gmx.at
Tue Mar 24 20:40:49 CET 2015


On Tue, Mar 24, 2015 at 11:26:36AM +0100, wm4 wrote:
> On Fri, 20 Mar 2015 18:33:37 +0100
> Michael Niedermayer <michaelni at gmx.at> wrote:
> 
> > On Fri, Mar 20, 2015 at 05:41:43PM +0100, wm4 wrote:
> > > The checks for the enums all were added in 2011 and 2012. They were
> > > likely for checking that Libav changes do not change the FFmpeg ABI.
> > > Given the amount of changes in the last 3 years, and the fact that
> > > barely any other ABI specifics are tested, keeping these tests makes
> > > not much sense anymore.
> > 
> > The tests check that ABI is not broken, for example if someone added
> > a entry in the middle of a enum it would break ABI to our own
> > previous release with same so-name.
> > its better if the assert punches the person who puts that in the
> > middle of a enum straight in the face instead of it being posted
> > to the mailing list and someone needing to ask for it to be fixed
> > and reposted with it moved to the end
> 
> Such invalid changes are easily caught in the patch reviews.

yes but its better if they are caught before the patch for the
review is even created


> 
> > Testing for this in a cleaner way is very welcome of course
> 
> Sure, but these barely test anything anyway and are not worth for the
> required uglification. You can just move them into some test program if
> you really want them.

yes i can but i have along todo list, and i will work on things
which appear more important to me first. That are for example bugs
some of which could be security relevant
So if its important to you that this gets moved soon then its
probably better if you move the code


[...]
> >
> > 
> > [...]
> > 
> > > -    av_assert0(LIBAVUTIL_VERSION_MICRO >= 100);
> > > -    av_assert0(HAVE_MMX2 == HAVE_MMXEXT);
> > 
> > These 2 lack any explanation of why they would be removed in the
> > commit message
> > 
> > [...]
> 
> Nobody would do anything that these assertions trigger, plus it's
> easily caught in patch review.

ive seen people break the MICRO >= 100 case
actually, theres no much ive not seen someone somewhere mess up

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

It is dangerous to be right in matters on which the established authorities
are wrong. -- Voltaire
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: Digital signature
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20150324/fa4ac7de/attachment.asc>


More information about the ffmpeg-devel mailing list