Loren Merritt <[email protected]> writes:

> On Wed, 8 Aug 2012, Mns Rullgrd wrote:
>> Diego Biurrun <[email protected]> writes:
>>
>>>  libavcodec/x86/Makefile                            |    6 +++---
>>>  libavcodec/x86/{dsputil_yasm.asm => dsputil.asm}   |    0
>>>  .../x86/{dsputilenc_yasm.asm => dsputilenc.asm}    |    0
>>>  libavcodec/x86/{vc1dsp_yasm.asm => vc1dsp.asm}     |    0
>>>  4 files changed, 3 insertions(+), 3 deletions(-)
>>>  rename libavcodec/x86/{dsputil_yasm.asm => dsputil.asm} (100%)
>>>  rename libavcodec/x86/{dsputilenc_yasm.asm => dsputilenc.asm} (100%)
>>>  rename libavcodec/x86/{vc1dsp_yasm.asm => vc1dsp.asm} (100%)
>>
>> Makes sense.  _yasm doesn't convey any information not already given by
>> the .asm suffix, given that all the x86 .asm files use yasm syntax.
>
> Same for _mmx.

In many of our cases, yes.  More generally, _mmx at least says
*something* that the rest of the filename does not.

> Do we have a standard for what to do when we have both a dsputil.c and a
> dsputil.asm, where the .o's would collide if we gave both their natural
> basename?

For ARM we use foo_init.c, foo.S, foo_armv6.S, etc.  That makes sense
there since the .c files never contain any actual code.

-- 
Måns Rullgård
[email protected]
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to