[FFmpeg-devel] [patch] Fix a warning in fft_mmx.asm on mac

[Michael Niedermayer]

On Tue, Apr 17, 2012 at 08:46:27AM -0700, Nico Weber wrote:
> Hi Michael,
> 
> thank you for your reply.
> 
> >> If you don't think this revised patch can be landed, please let me
> >> know what's missing.
> >
> > I think if there are still users around using old versions that need
> > the workaround then the workaround should stay and no configure check
> > should be added. (such check adds maintaince burden and its for a case
> > that wont get tested so breakage is not unlikely)
> >
> > OTOH if all users, all platforms and such are using new enough versions
> > then it just can be droped but
> >
> >> Yasm was fixed in its r2161 and yasm 0.8.0 (Apr 2010) contained this fix.
> >> Nasm was fixed in 2.06 (Jun 2009):
> >> https://groups.google.com/group/alt.lang.asm/browse_thread/thread/fcc85bbc3745d893
> >
> > Apr 2010 is not so long ago, i dont think this would have a zero effect
> 
> Sorry, I mistyped the year :-/ It's actually Apr 2009 (source:
> http://yasm.tortall.net/Download.html) Not sure if this changes much
> though.
> 
> > on our users. And i see nothing in your mail that would point to some
> > research that all users & platforms would be using a recent enough
> > version. But then this is huge work to do and the gain is fairly small
> 
> Note that this is a mac-only issue. People who build ffmpeg on mac
> need to download / build yasm manually (yasm isn't included with
> Xcode, and the nasm that is included doesn't pass the existing
> configure script), so if configure told them to upgrade to a
> less-than-three-years-old yasm, they would probably need to know what
> to do.
> 
> Is there any research I can do to estimate how many people would be
> affected by this change? What kind of data would you like to see that
> I can realistically collect?

with user installed sw iam not aware of any way to do it.

about the patch, it has another problem, it relies on "otool"
i dont know what that is but even if its standard on mac its not
standard on linux and yasm on linux seems to support to output macho
formats too, or at least they are listed as supported, ive not tried


> 
> > Also if you are concerned about warnings, there are many that can be
> > removed with less risk and troubble.
> 
> This is the last warning in all of chromium, including all third-party
> code. (Admittedly, we disabled several harmless warnings that ffmpeg
> doesn't seem to care about in the C parts. This yasm warning can't be
> disabled though.)

maybe instead of a yes or no to the removial of this code, it could
just be put under a __YASM_VERSION_ID__ based check ?
it must not break nasm of course
?

This would get rid of the warning and wouldnt break old yasm support


[...]
-- 
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/20120417/a65b074f/attachment.asc>