vorbis Log


Commit Date  
[ea6f75] by xiphmont xiphmont

Well. *that* deadlock bug has been lurking in seeking for a long time.
A bigger chunksize + just the right annodex file finally reproduced it
here.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13217 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-07-04 05:13:33 Tree
[321797] by xiphmont xiphmont

New vorbisfile multiplexing support passes built-in playback and seeking
tests.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13215 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-07-03 23:53:43 Tree
[fd1446] by xiphmont xiphmont

Multiplex support fixes; basic functions test OK on single stream files
now, time to test chains and multiplexing.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13214 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-07-03 21:31:52 Tree
[2e6eb9] by xiphmont xiphmont

Multiplexed stream support for vorbisfile *IN PROGRESS*
Don't use it, I know for certain it's broken.

Just a commit to make switching machines easier :-)

git-svn-id: http://svn.xiph.org/trunk/vorbis@13213 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-07-03 08:59:28 Tree
[1da9e7] by xiphmont xiphmont

Add return checks to all instances of seek_func and _seek_helper.
Should close bug #853.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13211 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-07-02 23:20:21 Tree
[6c7e9c] by xiphmont xiphmont

Although bug 552 was complaining about behavior that was illegal
anyway, added an additional proofing check to vorbisfile to aid a
developer figuring out what he's doing wrong.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13210 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-07-02 22:51:10 Tree
[d71f86] by xiphmont xiphmont

Fix for bug 577; used a different approach than the submitted patch
(rather than fudging an out of bound result after computation, do it
the right way to begin with)

Add test for fix to seeking_example.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13198 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-29 05:14:41 Tree
[431775] by xiphmont xiphmont

Draft filename changed, build system must be updated to match.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13197 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-29 05:07:49 Tree
[4d8830] by lu_zero lu_zero

06 commit to ietf

git-svn-id: http://svn.xiph.org/trunk/vorbis@13190 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-25 13:18:18 Tree
[04ee79] by xiphmont xiphmont

...if we're going to verify against minimum legal blocksize, we really
ought to verify against maximum too.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13179 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-23 06:42:39 Tree
[c5670a] by xiphmont xiphmont

More fixes to handle the null-entry codebook case. It appears the
failure to handle this case affects more of the code than thought;
many of the old low-bitrate crash cases trace back to this problem.

This also affects tremor. These changes are not yet thoroughly
tested.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13172 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-22 05:29:52 Tree
[d8d826] by xiphmont xiphmont

Apply patch as supplied in ticket 851

git-svn-id: http://svn.xiph.org/trunk/vorbis@13171 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-22 03:51:56 Tree
[836da0] by xiphmont xiphmont

The automated book training scripts managed to create a few residue
books with no codewords (in that mode with that partitioning scheme,
no values happened to fall into that partition). This is not really
an error.

However, the residue encoding code did not account for the possibility
of looking up a codebook best match and finding no entries. Because
the unused book only came up very rarely (four hours of training audio
didn't hit it once), it took a while to notice and only comes up on
platforms with subtly different FP than the machine used to do
training.

This bug affects very low bitrate stereo 8, 11, 16 and 22kHz modes;
added code to check for 'no match in codebook' (correct behavior is to
encode nothing) and for good measure, removed the four dummy codebooks
as there's no point in wasting a few bytes in the header to pack a
nothing.

Closes bug 1142.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13170 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-22 03:34:25 Tree
[25b184] by xiphmont xiphmont

Add additional check when attempting to encode values through
codebooks with no entries. Ideally that wouldn't happen, but it looks
like the automated codebook training ended up with a few 'null' books.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13169 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-22 03:01:47 Tree
[715e7c] by xiphmont xiphmont

Attempted fix for bug #300; the last comment identified a legitimate static bounds overrun that would have explained the behavior.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13168 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-22 01:02:19 Tree
[f8f7c5] by xiphmont xiphmont

Don't forget to test all changes in a code cleanup-- residue type 2 encoding was broken by last night's round of fixes.

Tested fixed.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13167 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-22 00:58:15 Tree
[5f0b14] by lu_zero lu_zero

Draft update

git-svn-id: http://svn.xiph.org/trunk/vorbis@13163 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-21 14:57:34 Tree
[875f58] by xiphmont xiphmont

Numerous small changes to resolve possible and certain issues found by
Coverity.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13162 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-21 07:06:25 Tree
[9a631b] by xiphmont xiphmont

Correct a cleanup dispatch error when a bitstream is rejected due to
illegal map/floor/residue type. The cleanup subdispatch was done by
type, unfortunately, the invalid type just read in. Fixed.

also bump lib revision.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13160 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-21 06:20:08 Tree
[ed73ba] by xiphmont xiphmont

Spec addition to clarify corner case; residue and floor decoded sections
may overflow the base vector's actual storage; this *must* be handled by
decoders.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13159 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-21 05:22:35 Tree
[8917ed] by xiphmont xiphmont

Catch one more omission in the vector length guarding

git-svn-id: http://svn.xiph.org/trunk/vorbis@13155 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-21 04:35:10 Tree
[604a1e] by xiphmont xiphmont

I had forgotten that the res2 top level interprets limits differently
than res0 and res1; guarding code (and garbled playback) fixed

git-svn-id: http://svn.xiph.org/trunk/vorbis@13154 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-21 04:22:33 Tree
[569985] by xiphmont xiphmont

If you're going to guard against max blocksize, make sure you choose the right blocksize.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13153 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-21 00:51:15 Tree
[6122c8] by xiphmont xiphmont

Additional bounds checking to guard against the case where a residue
or floor is decoding into a vector larger than the current blocksize.
This is not illegal, but no encoder has ever tried it and as such, it
had not been noticed that the code did not handle the case correctly.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13151 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-21 00:42:23 Tree
[00948a] by giles giles

Update change log with the news so far.

git-svn-id: http://svn.xiph.org/trunk/vorbis@13150 0101bb08-14d6-0310-b084-bc0e0c8e3800

2007-06-20 22:25:42 Tree
Older >