CVS change done by Oded Shimon CVS Update of /cvsroot/mplayer/main/DOCS/tech In directory mail:/var2/tmp/cvs-serv5137/DOCS/tech Modified Files: mpcf.txt Log Message: better max for frame code values, still allowing compact implementation Index: mpcf.txt =================================================================== RCS file: /cvsroot/mplayer/main/DOCS/tech/mpcf.txt,v retrieving revision 1.143 retrieving revision 1.144 diff -u -r1.143 -r1.144 --- mpcf.txt 12 Mar 2006 13:21:04 -0000 1.143 +++ mpcf.txt 12 Mar 2006 13:30:45 -0000 1.144 @@ -516,20 +516,20 @@ its pts differs by more then max_pts_distance from the last frame stream_id_plus1[frame_code] - must be <250 + MUST be <256 if it is 0, then the stream_id is coded in the frame data_size_mul[frame_code] - must be <16384 + MUST be <65536 data_size_lsb[frame_code] - must be <16384 + MUST be <65536 pts_delta[frame_code] - must be <16384 and >-16384 + MUST be <32768 and >-32768 reserved_count[frame_code] - must be <256 + MUST be <256 data_size data_size= data_size_lsb + data_size_msb*data_size_mul;
Hi On Sun, Mar 12, 2006 at 02:30:48PM +0100, Oded Shimon CVS wrote:
CVS change done by Oded Shimon CVS
Update of /cvsroot/mplayer/main/DOCS/tech In directory mail:/var2/tmp/cvs-serv5137/DOCS/tech
Modified Files: mpcf.txt Log Message: better max for frame code values, still allowing compact implementation
iam against this change, having a few reserved values 1. allows us to use these values in the future (using exactly 8 or 16 bit means that hw demuxers might not be easily able to be extended) 2. allows implementations to internally use the unused cases for flags and special cases 3. the extended values added are not usefull in practice 3A. you have 254 frame codes, some are needed for "exceptional" cases so >250 streams will hardly be hardcodeable in reality 3B. data_size_mul/lsb > 256 make little >16384 make no sense [...] -- Michael
On Sun, Mar 12, 2006 at 03:56:59PM +0100, Michael Niedermayer wrote:
Hi
On Sun, Mar 12, 2006 at 02:30:48PM +0100, Oded Shimon CVS wrote:
CVS change done by Oded Shimon CVS
Update of /cvsroot/mplayer/main/DOCS/tech In directory mail:/var2/tmp/cvs-serv5137/DOCS/tech
Modified Files: mpcf.txt Log Message: better max for frame code values, still allowing compact implementation
iam against this change, having a few reserved values 1. allows us to use these values in the future (using exactly 8 or 16 bit means that hw demuxers might not be easily able to be extended) 2. allows implementations to internally use the unused cases for flags and special cases 3. the extended values added are not usefull in practice 3A. you have 254 frame codes, some are needed for "exceptional" cases so >250 streams will hardly be hardcodeable in reality 3B. data_size_mul/lsb > 256 make little >16384 make no sense
OK, I'm not against reverting then. - ods15
On Sun, Mar 12, 2006 at 06:43:13PM +0200, Oded Shimon wrote:
On Sun, Mar 12, 2006 at 03:56:59PM +0100, Michael Niedermayer wrote:
Hi
On Sun, Mar 12, 2006 at 02:30:48PM +0100, Oded Shimon CVS wrote:
CVS change done by Oded Shimon CVS
Update of /cvsroot/mplayer/main/DOCS/tech In directory mail:/var2/tmp/cvs-serv5137/DOCS/tech
Modified Files: mpcf.txt Log Message: better max for frame code values, still allowing compact implementation
iam against this change, having a few reserved values 1. allows us to use these values in the future (using exactly 8 or 16 bit means that hw demuxers might not be easily able to be extended) 2. allows implementations to internally use the unused cases for flags and special cases 3. the extended values added are not usefull in practice 3A. you have 254 frame codes, some are needed for "exceptional" cases so >250 streams will hardly be hardcodeable in reality 3B. data_size_mul/lsb > 256 make little >16384 make no sense
BTW, regarding that, they are useful for storing mp3 packets in a single byte of overhead... - ods15
Hi On Sun, Mar 12, 2006 at 06:54:21PM +0200, Oded Shimon wrote:
On Sun, Mar 12, 2006 at 06:43:13PM +0200, Oded Shimon wrote:
On Sun, Mar 12, 2006 at 03:56:59PM +0100, Michael Niedermayer wrote:
Hi
On Sun, Mar 12, 2006 at 02:30:48PM +0100, Oded Shimon CVS wrote:
CVS change done by Oded Shimon CVS
Update of /cvsroot/mplayer/main/DOCS/tech In directory mail:/var2/tmp/cvs-serv5137/DOCS/tech
Modified Files: mpcf.txt Log Message: better max for frame code values, still allowing compact implementation
iam against this change, having a few reserved values 1. allows us to use these values in the future (using exactly 8 or 16 bit means that hw demuxers might not be easily able to be extended) 2. allows implementations to internally use the unused cases for flags and special cases 3. the extended values added are not usefull in practice 3A. you have 254 frame codes, some are needed for "exceptional" cases so >250 streams will hardly be hardcodeable in reality 3B. data_size_mul/lsb > 256 make little >16384 make no sense
BTW, regarding that, they are useful for storing mp3 packets in a single byte of overhead...
mp3 doesnt have packets > 16384 bytes [...] -- Michael
On Sun, Mar 12, 2006 at 10:41:43PM +0100, Michael Niedermayer wrote:
Hi
On Sun, Mar 12, 2006 at 06:54:21PM +0200, Oded Shimon wrote:
On Sun, Mar 12, 2006 at 06:43:13PM +0200, Oded Shimon wrote:
On Sun, Mar 12, 2006 at 03:56:59PM +0100, Michael Niedermayer wrote:
Hi
On Sun, Mar 12, 2006 at 02:30:48PM +0100, Oded Shimon CVS wrote:
CVS change done by Oded Shimon CVS
Update of /cvsroot/mplayer/main/DOCS/tech In directory mail:/var2/tmp/cvs-serv5137/DOCS/tech
Modified Files: mpcf.txt Log Message: better max for frame code values, still allowing compact implementation
iam against this change, having a few reserved values 1. allows us to use these values in the future (using exactly 8 or 16 bit means that hw demuxers might not be easily able to be extended) 2. allows implementations to internally use the unused cases for flags and special cases 3. the extended values added are not usefull in practice 3A. you have 254 frame codes, some are needed for "exceptional" cases so >250 streams will hardly be hardcodeable in reality 3B. data_size_mul/lsb > 256 make little >16384 make no sense
BTW, regarding that, they are useful for storing mp3 packets in a single byte of overhead...
mp3 doesnt have packets > 16384 bytes
He probably meant the >256 case. Rich
On Sun, Mar 12, 2006 at 02:30:48PM +0100, Oded Shimon CVS wrote:
CVS change done by Oded Shimon CVS
Update of /cvsroot/mplayer/main/DOCS/tech In directory mail:/var2/tmp/cvs-serv5137/DOCS/tech
Modified Files: mpcf.txt Log Message: better max for frame code values, still allowing compact implementation
This is wrong!!
Index: mpcf.txt =================================================================== RCS file: /cvsroot/mplayer/main/DOCS/tech/mpcf.txt,v retrieving revision 1.143 retrieving revision 1.144 diff -u -r1.143 -r1.144 --- mpcf.txt 12 Mar 2006 13:21:04 -0000 1.143 +++ mpcf.txt 12 Mar 2006 13:30:45 -0000 1.144 @@ -516,20 +516,20 @@ its pts differs by more then max_pts_distance from the last frame
stream_id_plus1[frame_code] - must be <250 + MUST be <256 if it is 0, then the stream_id is coded in the frame
data_size_mul[frame_code] - must be <16384 + MUST be <65536
data_size_lsb[frame_code] - must be <16384 + MUST be <65536
This has nothing to do with making framecode table compact. It's about storage efficiency.
pts_delta[frame_code] - must be <16384 and >-16384 + MUST be <32768 and >-32768
I suppose this is ok, I dunno. Rich
participants (4)
-
Michael Niedermayer -
Oded Shimon -
Rich Felker -
syncmail＠mplayerhq.hu