[Ffmpeg-devel] [PATCH] BMP encoder
Michel Bardiaux
mbardiaux
Tue Oct 31 14:33:24 CET 2006
Michael Niedermayer wrote:
> Hi
>
> you are quicker with sending patches then iam with thinking about how to
> optimally solve the issues ...
>
> lets see we have
> LD*/ST* from dsputil.h which read and write unaligned native endian
> LE_*/BE_* from avcodec.h which read unaligned little and big endian
> bytestream_get_* in bytestream which read unaligned little endian and
> increase a pointer
>
> IMHO bytestream.h should only contain functions which increase a pointer
> similar to what is currently in there
OK, no touchy bytestream.h. But I dont understand why you passed the
introduction (and use in snow) of 2 different APIs for essentially the
same thing.
>
> macros to read and write little,big and native endian should be IMHO in a
> seperate file, also LE*/BE*/LD*/ST* should be moved into that
Something not clear to me here: LE*/BE*/LD*/ST* to, say, bytebuffer.h
(OK?), that's clear. But apart from these, which "macros to read and
write little,big and native endian"?
> and optionally
> be cleaned up in the sense that native endian on BE should be equal to BE_*
> and similarly on LE
> furthermore we could rename them so their names are more consistent
> something like:
> LDLE_XY load little endian XY bit
> STLE_XY store little endian XY bit
> LDBE_XY load big endian XY bit
> STBE_XY store big endian XY bit
> LD_XY load native endian XY bit
> ST_XY store native endian XY bit
>
> or we could leave their names and just add the missing 2
> LE_XY load little endian XY bit
> STLE_XY store little endian XY bit
> BE_XY load big endian XY bit
> STBE_XY store big endian XY bit
> LDXY load native endian XY bit
> STXY store native endian XY bit
>
> comments?
I think I'dd the the first, more symmetric.
>
> also note the bmp encoder just depends on moving ST* into bytesometing.h
> (something!=stream) or writing STLE_* the rest is a seperate issue
>
> [...]
>> @@ -22,6 +22,8 @@
>> #ifndef FFMPEG_BYTESTREAM_H
>> #define FFMPEG_BYTESTREAM_H
>>
>> +// These assume alignment
>
> they do not
I have posted a correction.
>
>
> [...]
--
Michel Bardiaux
R&D Director
T +32 [0] 2 790 29 41
F +32 [0] 2 790 29 02
E mailto:mbardiaux at mediaxim.be
Mediaxim NV/SA
Vorstlaan 191 Boulevard du Souverain
Brussel 1160 Bruxelles
http://www.mediaxim.com/
More information about the ffmpeg-devel
mailing list