[FFmpeg-devel] [PATCH] Bunch of accumulated patches...

[Sigbjørn Skjæret]

[..Please CC me if you want me to reply, otherwise I'll have to break the thread like now..]

@michael:
>> av_fast_realloc.diffs: Since all the other mem-funcs have been moved to avutil
>> this one really belongs there as well...
> this is no reason to make it part of the public API of libavutil

Uhm, why not? It's part of avcodec's public API, just like the other funcs were
before they were moved, and it's just a variation of av_realloc anyway.

>> averror.diffs changes AVERROR_NOFMT to use a commonly available define instead
>> (arbitrarily chosen, anything would do), however as this might be viewed as
>> breaking ABI there's also averror2.diffs which will define EILSEQ if it doesn't
>> exist already. Pick the one you like the best...
> libav* cannot #define standard E* in a public header
> also arbitrarily choosen E* as error codes is not acceptable

Ok, then I'm open to suggestions, what is the best suited errorcode? EILSEQ
doesn't define the error any better than ENOEXEC...

>> rationaldecls.diffs: Enclose directly applied AVRationals in parenthesis to
>> avoid preprocessing issues (ok, this one is really specific to my build since I
>> have defines for those funcs, but it doesn't hurt anyway and perhaps increases
>> readability slightly).
> fix whatever you do with the preprocessor

Well, I can't, the problem is really that the functions are defined as macros
here, and the preprocessor thus balks on the "," thinking there are too many
args, enclosing the whole casted struct fixes that, and is slightly more
readable IMO .. but if you don't want to apply that patch it's only my problem,
so no big deal. ;)

@Luca:
>> broken_mergel.diffs: This is a regression-fix, fixes erroneously (and broken)
>> applied defines when compiling with AltiVec-enabled GCC2.
> how broken?

Completely.

The AltiVec-enabled GCC2 uses the same vector-declaration method as Apple's
GCC, ie () instead of {}.

@Luca&Michael:
>> IMHO its a waste of time unless you have absolutely nothing else to do
>> AFAIK noone uses gcc2 on ppc?
> Nobody that I know of.

I know of plenty of ppl who do (and not just PPC), but that's beside the point.


- CISC