svn commit: r375696 - head/multimedia/libx264

[Kubilay Kocak]

On 28/12/2014 9:01 PM, Gerald Pfeifer wrote:
> On Sunday 2014-12-28 04:59, Kubilay Kocak wrote:
>> Log:
>>   multimedia/libx264: Fix PGO/GCC options
>>   
>>   - Revert options helpers for PGO and GCC .. *shakes fist*. We end up
>>     setting USE_GCC=yes yes otherwise, and that's not valid.
>>   - libx264 and x264 share the same distfiles, use the same DIST_SUBDIR
> 
>> +.if ${PORT_OPTIONS:MGCC}
>> +USE_GCC=	yes
>> +.elif ${PORT_OPTIONS:MPGO}
>> +USE_GCC=	any
>> +.endif
> 
> Why not USE_GCC=yes in either case?
> 
> Using anything like the old version of GCC in base (GCC 4.2) for
> anyone asking for PGO feels just, hmm, odd.  Plus it complicates
> the testing matrix.
> 
> Gerald
> 

'any' wont clobber user values with the latest or default version as
detailed in Mk/bsd.gcc.mk, according to my reading and `make test-gcc`
for each of the permutations.

In this case I prefer to preserve user choice, and I can't guess what
and how users want to do with the software.

Speaking from personal experience, I would not want > 1 gcc versions
built and/or installed if I was forced for whatever reason to use it
(eg: PGO until libprofile_rt in Clang lands in FreeBSD), whether or not
4.2.1 provided statistically significant benefits over later versions.

tl;dr I'm thinking of the users. I hope that's good enough :)

Koobs