[PATCH] uint_fast64_t in configure when int_fastXY_t failed
Synchronized with libavutil/common.h, fixes compilation on Sparc. -- Zuxy Beauty is truth, While truth is beauty. PGP KeyID: E8555ED6
Hi, Zuxy Meng wrote:
Synchronized with libavutil/common.h, fixes compilation on Sparc.
------------------------------------------------------------------------
Index: configure =================================================================== --- configure £¨ÐÞ¶©°æ 19431£© +++ configure £¨¹¤×÷¿½±´£© @@ -2786,7 +2786,8 @@ typedef signed int int_fast32_t; typedef unsigned char uint_fast8_t; typedef unsigned int uint_fast16_t; - typedef unsigned int uint_fast32_t;' + typedef unsigned int uint_fast32_t; + typedef unsigned long long uint_fast64_t;' fi echores "$_fast_inttypes"
Looks good to me. Unless Diego disagrees or apply it himself, I'll commit it tomorrow. Guillaume
Hi, Guillaume POIRIER wrote: [..]
typedef unsigned int uint_fast16_t; - typedef unsigned int uint_fast32_t;' + typedef unsigned int uint_fast32_t; + typedef unsigned long long uint_fast64_t;' fi echores "$_fast_inttypes"
Looks good to me. Unless Diego disagrees or apply it himself, I'll commit it tomorrow.
BTW, a quick grep revels that uint64_t is only used in imported FFmpeg's libs: ./libavutil/common.h:typedef uint64_t uint_fast64_t; ./libavcodec/vorbis.c: temp2=(((uint_fast64_t)temp) * inverse_class)>>32; So it's normal that the absence of uint64_t in the configure script went unnoticed. Guillaume
On Fri, Aug 18, 2006 at 01:25:39PM +0200, Guillaume POIRIER wrote:
Guillaume POIRIER wrote:
[..]
typedef unsigned int uint_fast16_t; - typedef unsigned int uint_fast32_t;' + typedef unsigned int uint_fast32_t; + typedef unsigned long long uint_fast64_t;' fi echores "$_fast_inttypes"
Looks good to me. Unless Diego disagrees or apply it himself, I'll commit it tomorrow.
BTW, a quick grep revels that uint64_t is only used in imported FFmpeg's libs:
./libavutil/common.h:typedef uint64_t uint_fast64_t; ./libavcodec/vorbis.c: temp2=(((uint_fast64_t)temp) * inverse_class)>>32;
So it's normal that the absence of uint64_t in the configure script went unnoticed.
And looking closer it seems that [u]int_fast* seems to be used in libmpeg2/idct_alpha.c and libavcodec only. Maybe there's a way to get rid of it completely outside libavcodec. Diego
On Fri, Aug 18, 2006 at 02:39:22PM +0800, Zuxy Meng wrote:
Synchronized with libavutil/common.h, fixes compilation on Sparc.
Hmmm
--- configure ???????? 19431?? +++ configure ???????????? @@ -2786,7 +2786,8 @@ typedef signed int int_fast32_t; typedef unsigned char uint_fast8_t; typedef unsigned int uint_fast16_t; - typedef unsigned int uint_fast32_t;' + typedef unsigned int uint_fast32_t; + typedef unsigned long long uint_fast64_t;'
In libavutil/common.h there is a signed variant as well. And why not switch to that header outright? libavutil is now mandatory for MPlayer anyway. Diego
On Fri, Aug 18, 2006 at 01:36:12PM +0200, Diego Biurrun wrote:
On Fri, Aug 18, 2006 at 02:39:22PM +0800, Zuxy Meng wrote:
Synchronized with libavutil/common.h, fixes compilation on Sparc.
Hmmm
--- configure ???????? 19431?? +++ configure ???????????? @@ -2786,7 +2786,8 @@ typedef signed int int_fast32_t; typedef unsigned char uint_fast8_t; typedef unsigned int uint_fast16_t; - typedef unsigned int uint_fast32_t;' + typedef unsigned int uint_fast32_t; + typedef unsigned long long uint_fast64_t;'
In libavutil/common.h there is a signed variant as well.
And why not switch to that header outright? libavutil is now mandatory for MPlayer anyway.
Anyway, the patch is small and harmless enough that I've applied it anyway, looking for better solutions nonwithstanding. Diego
participants (3)
-
Diego Biurrun -
Guillaume POIRIER -
Zuxy Meng