Commit graph

2789 commits

Author SHA1 Message Date
Karl Blomster
d56982ed4a Move the SMPTE timecode stuff out of the AssTime class and put it in a class of its own, FractionalTime. The AssTime to SMPTE part is tested and verified to work (in fact it looks like I unintentionally fixed an overlap bug in the TranStation export filter), while the SMPTE to AssTime part is completely untested (but it's currently not used anywhere so it's not like it matters).
Originally committed to SVN as r2909.
2009-05-10 03:50:58 +00:00
Karl Blomster
179d23fd5c Fix a memleak in the FFMS2 video provider (the frame was never unallocated on video close). Patch by Harukalover, fixes #838
Originally committed to SVN as r2908.
2009-05-10 00:12:04 +00:00
Fredrik Mellbin
2cab187fd8 FFMS2: beta 8
Originally committed to SVN as r2907.
2009-05-09 21:59:24 +00:00
Karl Blomster
3bdbcd7b5e Fix a minor bug that caused the add leadin/leadout buttons to not update the time edit boxes. Patch by Harukalover, fixes #836.
Originally committed to SVN as r2906.
2009-05-07 17:52:29 +00:00
Karl Blomster
87a39833b8 The buffer length/wanted latency in the dsound2 audio player are now tweakable in config.dat; the options are called "audio dsound buffer latency" and "audio dsound buffer length" respectively. Defaults to 100/5 respectively (should work for the people who used to get buffer errors); for better responsiveness try 80/4.
Originally committed to SVN as r2905.
2009-05-07 14:47:36 +00:00
Karl Blomster
bfe40a6667 Fix a bug (in a somewhat questionable manner) that would cause the text file writer to occasionally think the system locale was Unicode when it wasn't (by using an uninitialized variable in a condition). Should fix the issue with the SRT export filter failing to write "1" on the first line when using "local" as the text encoding.
Originally committed to SVN as r2904.
2009-05-06 19:14:10 +00:00
Karl Blomster
c6fe68c76d Add Aegisub::Script::set_undo_point() to auto4-perl. Currently not all that useful since auto4-perl currently only supports macros and nothing else, and using a macro sets an undo point after execution automatically anyway.
Originally committed to SVN as r2903.
2009-05-06 12:25:08 +00:00
Karl Blomster
02f7d0006c change how ffmsindex is linked to ffms2 (use a #pragma instead of an extra dependency in the project file)
Originally committed to SVN as r2902.
2009-05-05 21:43:49 +00:00
Niels Martin Hansen
36dbd86dd6 Oops, had swapped size and value parameters in a memset call.
Originally committed to SVN as r2901.
2009-05-04 19:15:57 +00:00
Amar Takhar
c9affe9aeb Sigh, make gawk happy, I despise gawk someone please get rid of it forever.
Originally committed to SVN as r2900.
2009-05-04 08:06:26 +00:00
Amar Takhar
8c8a0abe38 Add a note why the 'sed' hack is nessicary.
Originally committed to SVN as r2899.
2009-05-04 06:52:45 +00:00
Amar Takhar
f893e8a29f Add a crappy sed hack to silence gcc warnings, sigh, I wish ImageMagick would update convert already.
Originally committed to SVN as r2898.
2009-05-04 06:51:02 +00:00
Amar Takhar
5b356f2755 s/static/const/ to silence gcc warnings.
Originally committed to SVN as r2897.
2009-05-04 06:50:37 +00:00
Amar Takhar
2090994256 Silence a bunch of gcc 'newline' warnings.
Originally committed to SVN as r2896.
2009-05-04 06:50:12 +00:00
Amar Takhar
b00e485149 Plug a bunch of memory leaks from uau in the mplayer2 repo: http://repo.or.cz/w/mplayer.git?a=commit;h=312d9e4b104741b834aa5d71b02228d0cd988a4e
libass: Fix cache lookup problem causing memory bloat

The cache code did hash lookups by storing key values in struct fields
and then hashing and comparing the struct as a single memory block. In
at least one case such a struct contained uninitialized padding bytes
which prevented the complete memory area of the struct from matching
even though the fields did. As a result the code failed to find
existing objects in the cache and stored new versions of them, causing
gigabytes of memory use in some circumstances. Initializing the struct
memory to zero before writing the fields avoided such memory use in
tests but is not guaranteed to work if I interpret the C standard
correctly (the compiler is allowed to write garbage over padding bytes
when changing struct member values).

Change the code to use struct-specific hashing and comparison
functions that work field by field to guarantee correct behavior.
Create these by replacing the struct definition with a template that
lists the fields and can be used the generate each of struct
definition, hash function and compare function with some preprocessor
magic (otherwise every field would need to be listed separately in all
three).

Originally committed to SVN as r2895.
2009-05-04 06:48:21 +00:00
Fredrik Mellbin
0a221706d5 FFMS2:
Fix an old bug that would drop the first or second video frame
Greatly improved audio decoding by adding a small internal cache

Originally committed to SVN as r2894.
2009-05-03 19:25:54 +00:00
Karl Blomster
869e5c7594 Implemented an automated cache cleaner that keeps the FFMS2 index cache at a manageable size. Cleaning is started when video is opened with the FFMS2 video provider and runs in its own thread. Removal is based on last access time, and the FFMS2 providers now explicitly touch the index files on reading to make sure access time is updated.
By default the cache is kept below 42MB and 20 files; to change the defaults change "FFmpegSource max cache size" and "FFmpegSource max cache files" in config.dat. Either or both may be set to 0, but the cleaner will not delete the last index file in the directory, regardless of cache settings and of the file's size.

Originally committed to SVN as r2893.
2009-05-03 18:05:30 +00:00
Karl Blomster
e3ce74248d Fix compilation breakage introduced in r2891.
Originally committed to SVN as r2892.
2009-05-02 20:06:06 +00:00
Karl Blomster
45627548d6 Don't let the MKV parser crash on certain MKV's for which it cannot determine the filesize by itself. (Adaption of Myrsloik's similar fix for FFMS2.)
Originally committed to SVN as r2891.
2009-05-02 19:40:55 +00:00
Fredrik Mellbin
487dc50409 FFMS2: Add a missing function the updated matroska parser expects
Originally committed to SVN as r2890.
2009-05-01 16:58:15 +00:00
Amar Takhar
1cbd29531c Add an sh script to generate snap tarballs upon successful builds of UNIX platforms (excluding Darwin).
Originally committed to SVN as r2889.
2009-05-01 03:58:23 +00:00
Niels Martin Hansen
b840c9f97e Ensure some fields in VideoContext get initialised.
Originally committed to SVN as r2888.
2009-04-30 22:13:10 +00:00
Karl Blomster
4445d812ce Re-update the license on Haali's matroska parser.
Originally committed to SVN as r2887.
2009-04-29 21:19:45 +00:00
Amar Takhar
a9e1f80c8c Update libass from MPlayer r29129: Don't automatically wrap lines when WrapStyle==2.
Originally committed to SVN as r2886.
2009-04-29 21:06:54 +00:00
Fredrik Mellbin
fad9a58240 FFMS2: Update Haali's matroska parser
Originally committed to SVN as r2885.
2009-04-29 21:06:07 +00:00
Karl Blomster
2f7b62d206 Update Haali's MatroskaParser.c/.h in the main source tree (there's a copy in the FFMS2 folder as well, which will be updated later). Should fix a crash with x264-generated MKV's.
Originally committed to SVN as r2884.
2009-04-29 21:00:08 +00:00
Niels Martin Hansen
f0b442f4b4 Let AudioProvider::GetAudioWithVolume() catch exceptions in AudioProvider::GetAudio() and return blank audio when that happens. Errors are not logged or otherwise handled. This fixes #800 for most cases, even if poorly.
Originally committed to SVN as r2883.
2009-04-29 19:30:02 +00:00
Niels Martin Hansen
a2822356bf Some more delayload vsprops files.
Originally committed to SVN as r2882.
2009-04-29 19:08:40 +00:00
Karl Blomster
1f062a5553 remove duplicate variable
Originally committed to SVN as r2881.
2009-04-29 17:42:48 +00:00
Karl Blomster
9c93aeb179 Fix a few memory leaks in the FFMS2 providers. Patch by Myrsloik.
Originally committed to SVN as r2880.
2009-04-29 17:40:02 +00:00
Fredrik Mellbin
27e90e973f FFMS2: Beta 7 release
Fix 3 gcc warnings

Originally committed to SVN as r2879.
2009-04-29 16:39:15 +00:00
Karl Blomster
8bc3a39d0c Don't fail to read subtitles from MKV's that are named .mks or .mka, both of those are reported as supported formats in the "open subtitles" dialog box.
Originally committed to SVN as r2878.
2009-04-29 14:04:23 +00:00
Amar Takhar
29721b983e s/32/??/ to include 64bit builds (what was I thinking?).. it's better than using '*' which could lead to a disaster.
Originally committed to SVN as r2877.
2009-04-29 08:07:22 +00:00
Amar Takhar
1fb14af113 Use 'SetFile' instead of '/usr/bin/SetFile' when creating the .dmg, 10.5 puts it in /usr/bin, on 10.4 it's in /Developer/Tools/.
Originally committed to SVN as r2876.
2009-04-29 07:55:40 +00:00
Amar Takhar
05af876c1d * Favour '/usr/bin/perl' over 'perl' to ensure the base perl is always used. (it's required)
Originally committed to SVN as r2875.
2009-04-29 06:50:08 +00:00
Amar Takhar
02210e3669 * Re-format the arch detection case.
* Fix an annoying buy, I hate the loops configure gets stuck in when you do "$blah" = "$blah" vs test "$blah..

Originally committed to SVN as r2874.
2009-04-29 00:49:38 +00:00
Amar Takhar
d769707641 * Add support in configure.in to detect the arch.
* Add two new targets, osx-tinderbox-(bundle|dmg) to handle tinderbox building.

Originally committed to SVN as r2873.
2009-04-28 23:25:42 +00:00
Amar Takhar
3267c11fed Change the buildslave dist name to dist.7z so the waterfall isn't cluttered. There's a link to the archive so it won't be an issue.
Originally committed to SVN as r2872.
2009-04-28 17:59:03 +00:00
Karl Blomster
876f726a6d Don't try to synch audio when there is no audio player/provider present. (Could cause a crash under certain rare conditions.) Thanks Avi for report and fix.
Originally committed to SVN as r2871.
2009-04-28 10:23:20 +00:00
Amar Takhar
800e57897b Add a batch file to generate a distfile that can be uploaded.
Originally committed to SVN as r2870.
2009-04-28 07:04:53 +00:00
Amar Takhar
9774fbb1c8 Remove an extra line (more testing.)
Originally committed to SVN as r2869.
2009-04-28 05:25:37 +00:00
Amar Takhar
77643ae596 Remove some EOF whitespace (more testing.)
Originally committed to SVN as r2868.
2009-04-28 05:24:28 +00:00
Amar Takhar
9079f1576f Remove some EOF whitespace.
Originally committed to SVN as r2867.
2009-04-28 05:21:23 +00:00
Amar Takhar
8a13a78229 Add some error-checking and exit 1's.
Originally committed to SVN as r2866.
2009-04-28 05:17:33 +00:00
Amar Takhar
d929e60b4c Add exit codes to various commands to ensure the script fails and doesn't proceed to the next step(s).
Originally committed to SVN as r2865.
2009-04-28 05:11:32 +00:00
Amar Takhar
501782d605 Testing the tinderbox (a stub commit but still useful.)
Originally committed to SVN as r2864.
2009-04-28 05:08:18 +00:00
Amar Takhar
bf99cbcd58 Testing the tinderbox.
Originally committed to SVN as r2863.
2009-04-27 20:58:42 +00:00
Niels Martin Hansen
03ec2ae033 Fix include file search path in CSRI project file for VC++ in x64 configurations.
Originally committed to SVN as r2862.
2009-04-26 21:01:18 +00:00
Karl Blomster
9444ebd0d1 Fix some more minor editbox regressions introduced in r2859. Patch by Harukalover, updates #794
Originally committed to SVN as r2861.
2009-04-26 04:31:19 +00:00
Karl Blomster
2ba20b12c8 Fix a minor graphical glitch in the time edit boxes introduced in r2859. Patch by Harukalover. updates #794
Originally committed to SVN as r2860.
2009-04-26 01:23:01 +00:00