[Ffmpeg-cvslog] CVS: ffmpeg/libavcodec utils.c, 1.173, 1.174 avcodec.h, 1.455, 1.456

[Rich Felker]

On Tue, Feb 21, 2006 at 03:00:41PM -0800, Mike Melanson wrote:
> Alex Beregszaszi wrote:
> >Hi,
> >
> >
> >>>@@ -266,11 +266,14 @@
> >>>};
> >>>
> >>>/* currently unused, may be used if 24/32 bits samples ever
> >>>supported */
> >>>+/* all in native endian */
> >>>enum SampleFormat {
> >>>-    SAMPLE_FMT_S16 = 0,         ///< signed 16 bits
> >>>+    SAMPLT_FMT_NONE = -1,
> >>>+    SAMPLE_FMT_U8,              ///< unsigned 8 bits
> >>>+    SAMPLE_FMT_S16,             ///< signed 16 bits
> >>>+    SAMPLE_FMT_S24,             ///< signed 24 bits
> >>>    SAMPLE_FMT_S32,             ///< signed 32 bits
> >>>    SAMPLE_FMT_FLT,             ///< float
> >>>-    SAMPLE_FMT_DBL,             ///< double
> >>>};
> >>
> >>I guess this breaks the ABI (and API as you remove SAMPLE_FMT_DBL)
> >>so this certainly deserve a version number increase.
> >
> >
> >It was never used. Michael, can you comment?
> 
> 	If it was never used, this might be a good opportunity to change 
> 	_FLT and _DBL to something more consistent and portable such as _F32 and 
> _F64. Just an idea.

AFAIK _FLT and _DBL are portable to any IEEE-float-conformant
compiler. These sizes are set in stone, not subject to change, unlike
integer sizes.

Rich