[MPlayer-dev-eng] [PATCH] two independent patches for configure

[Alexander Strasser]

Xidorn Quan wrote:
> On Sat, Oct 13, 2012 at 6:29 PM, Alexander Strasser <eclipse7 at gmx.net>wrote:
> > Xidorn Quan wrote:
> > > On Tue, Oct 9, 2012 at 10:54 PM, Xidorn Quan <quanxunzhen at gmail.com>
> > wrote:
> > > > Two independent patches for configure are attached.
> > > >
> > > > The first patch adds more cpu type detection. It can now recongize
> > > > atom, corei7, and corei7-avx. References:
> > > > http://en.gentoo-wiki.com/wiki/Safe_Cflags/Intel
> > > > http://software.intel.com/en-us/comment/1137881
> > > > http://www.acsu.buffalo.edu/~charngda/x86cpu.html
> > > >
> > > > The second patch moves some code which dependent on information of cc
> > > > down so that it can work properly.
> > > >
> > >
> > > Ping. These two patches are simple.
> > >
> > > The first patch can improve performance of mplayer when compiled by
> > > compilers doesn't support cpu type autodetecting, like clang, and
> > > runtime cpu detection is disabled. I have given the references for
> > > such detections.
> > >
> > > The second patch removes unused flags for clang to silent some
> > > compiling warning.
> >
> >   I apologize for my slowness. Please be a bit more patient. I
> > will look into your patches - as my rather little time permits.
> >
> >   Currently I am looking at your other patch.
> 
> 
> Well, there have been about 10 days since you applied my other patch.
> Do you now have time to look into these patches? The patches has been
> here for more than 2 weeks.

  In absence of any further comments on these patches:

  I am in favor of your 2nd patch which moves the extra_cflags
modification after compiler detection. I plan to commit it in
one to two days.

  I still have a few questions though: Is clang invoked via cc
program name on your installation? Could you test if it still
works as intended with gcc on darwin?

  On the first patch. I am not against it and it mostly looks
fine. I still need to check the facts though. On the critique
of wm4, I do not think it makes sense to reject this patch
particularly. I also don't buy the packaging argument and still
if I am wrong the setting-of-compiler-switches-for-arch-dependant-
optimization-feature needs to be ripped off all together.

  Comments welcome...

  Alexander