2006-05-07    <mschimek@users.sf.net>

	* common/structpr.pl: ILP64 fix.

2005-12-01    <mschimek@users.sf.net>

	* configure.in: O_LARGEFILE check was missing.

2005-12-30    <mschimek@users.sf.net>

	* devices/v4l25.c: Request V4L2_PRIORITY_RECORD.

2005-10-25    <mschimek@users.sf.net>

	* configure.in: Added AM_MAINTAINER_MODE.
	* video/Makefile.am: Build BUILT_SOURCES only in maintainer mode,
	  distribute BUILT_SOURCES.
	* common/Makefile.am: Build BUILT_SOURCES only in maintainer mode.

2005-10-20    <mschimek@users.sf.net>

	* main.c (break_sequence, main), systems/output.c (output_stdout):
	  Open with O_LARGEFILE only if O_LARGEFILE is supported by libc.

2005-09-03    <mschimek@users.sf.net>

	* devices/v4l25.c: Didn't reset cropping window and limit image
	  size to active video size.

2005-08-15    <mschimek@users.sf.net>

	* b_mp1e.c: Added cut_output variable for an unused mp1e function.

2005-06-30    <mschimek@users.sf.net>

	* common/structpr.pl: Didn't log VIDIOC_G|S_STD.

2005-06-28    <mschimek@users.sf.net>

	* options.c (parse_option): GCC 4.0 rejects a static function
	  declaration in a function, moved that out.

2005-04-24    <mschimek@users.sf.net>

	* global_data.c, options.c (parse_option), options.h (options),
	  systems/output.c (output_stdout):
	  New option to cut output into sub-1GB segments. New option to
	  change number of capture buffers.

2005-04-11    <mschimek@users.sf.net>

	* devices/v4l25.c (v4l25_init): Change audio mode (with
	  v4l VIDIOCSAUDIO for compat. with broken drivers) matching
	  the -a option.
	* video/mpeg1.c (option_set), video/mpeg2.c (option_set):
	  Doubled upper limit of video bit rate.
	* configure.in: Bumped version number to 1.9.6.

2005-03-16    <mschimek@users.sf.net>

	* RTE release 0.5.6.

2005-03-16    <mschimek@users.sf.net>

	* b_mp1e.c (send_empty_ca): Do not return EOF buffer which contains
	  no data.  Caused an assert (c->dequeued == 0) when stopping mp1e
	  video recording in Zapping.
	* audio/mp2.c (parameters_set): Must align output buffer size for
	  bflush access.  SF bug #1162581 (make check segv due to buffer
	  overflow).
	* audio/mp2.c (parameters_set), video/mpeg2.c (parameters_set),
	  video/mpeg2.c (parameters_set): No copy of codec->params to self.

2005-02-28    <mschimek@users.sf.net>

	* RTE release 0.5.4.
	
2005-02-26    <mschimek@users.sf.net>

	* common/alloc.h, common/alloc.c: Don't use memalign/free, they're
	  unreliable.

2005-02-20    <mschimek@users.sf.net>

	* systems/systems.c (mpeg_header_name): Constness fix.
	* systems/vcd.c (schedule), systems/mpeg2.c (schedule),
	  systems/mpeg1.c (schedule): VCD mux segv at stream end because
	  dts in schedule() wrapped and an already finished stream was
	  scheduled for output again. Added similar check to mpeg1 and mpeg2
	  routines, just in case.
	* b_mp1e.c: gcc 4.x fixes.
	(set_output): Segv in rte_error_printf() due to missing parameter.
	(codec_set): mp1e_mpeg1_vcd and mpeg2_audio_layer2 do not combine.
	* common/fifo.c, common/types.h (asserts_fail): Constness fix.
	* systems/libsystems.h (mpeg_header_name), common/mmx.c (cpuid_t),
	  common/fifo.c (asserts_fail), common/types.h: Constness fixes.
	* audio/mp2.c: Lvalue cast fixes (gcc 4.x).
	(mp1e_mp2): asm parameter fix (gcc 4.x).
	* audio/mp2.c, common/bstream.h (bcatq): gcc 4.x fixes.
	* common/list.h (destroy_list): Suppress unused parameter warning.
	* common/mmx.h (swab32): Removed redundant const in return type.
	  (cpuid_t): Unsiged char fix.
	* video/mpeg2.c (uninit), video/mpeg1.c (uninit): Segv on FreeBSD
	  because the function called free() instead of free_aligned().
	* common/alloc.c, common/alloc.h (alloc_aligned): Fixed.

2004-01-19     <japie@deserver.nl>

	* common/mmx.c, common/mmx.h, main.c, mp1e.1,
          options.c, mpeg1.c, motion.c:
          Added Via Nehemiah (C3) CPU.

2004-05-19    <mschimek@users.sf.net>

	* common/mmx.h: Compiler identification update.
	* video/motion.h: Troubles with gcc 3.4 and attribute regparm,
	  removed.

2004-04-23    <mschimek@users.sf.net>

	* configure.in: Added check and warning about older version
	  in other directory.

2004-04-21    <mschimek@users.sf.net>

	* common/device.c, common/device.h, common/structpr.pl,
	  common/Makefile.am (libcommon_la_SOURCES), devices/v4l25.c:
	  Added ioctl logging functions.

2004-04-08	<mschimek@users.sf.net>

	* devices/v4l25.c: Fixed incomplete v4l2_buffer
	  initialization.

2003-03-06	<mschimek@users.sf.net>

	* video/video.h, video/mpeg1.c, video/filter_mmx.s,
	  video/filter.c, devices/v4l2.c, devices/v4l25.c:
	  Added UYVY support.
	
2003-02-10	<mschimek@users.sf.net>

	* common/videodev25.h: Updated.

2003-01-15	<mschimek@users.sf.net>

	* common/videodev25.h, devices/v4l25.c, main.c: Added
	  v4l2 (Linux 2.5) interface, untested.
	* devices/alsa.c: "New API" corrections.

2003-01-02	<mschimek@users.sf.net>

	* macros/alsa.m4: Updated from ALSA 0.9.0rc6.
	* configure.in: New check for ALSA 0.9.x.
	* devices/Makefile.am: Added ALSA_CFLAGS|_LIBS.
	* devices/alsa.c: Rewrote ALSA 0.9 interface, works now, with
	  proper timestamping. Removed ALSA 0.5 routines.
	* global_data.c: ALSA device name fix.

2002-12-25	<mschimek@users.sf.net>

	* systems/vcd.c: Fixed status.bytes_out bug.

2002-12-17	<mschimek@users.sf.net>

	* b_mp1e.c: VCD output buffer too small.
	* common/mmx.h: gcc minor version++.

2002-11-11	<zhubr@users.sf.net>

	* devices/v4l.c: Avoid grab width padding (Allows to use 
	  horizontal decimating modes on ATI radeons in PAL/SECAM);
	* devices/v4l.c: Try yet harder to automatically correct grab 
	  size before giving up;
	* devices/v4l.c: Better use of > 2 kernel buffers (if present);
	* devices/v4l.c: Fixed minor typos (hopefully);
	* devices/v4l.c: Changed ffmpeg's author obsolete alias to his 
	  actual name in the header.

2002-10-05	<mschimek@users.sf.net>

	* video/mpeg1.c, video/mpeg2.c: GOP header fix, incorrect
	  closed_gop flag.

2002-10-02	<mschimek@users.sf.net>

	* video/vlc_mmx.s: Correction.
	* video/ratectl.h: Splitted for future field interleaving.
	* video, audio, vbi, systems, common, devices, test: #include
	  config.h didn't work reliable, changed to Makefile.am
	  compiler flag.

2002-09-12	<mschimek@users.sf.net>

	* video: encode_intra now reentrant. Let's hope speed
	  didn't suffer too much.

2002-09-08	<mschimek@users.sf.net>

	* video/filter.c: 4:2:0 cropping fix by Nikolai Zhubr.
	* main.c: HAVE_PROGRAM_INVOCATION_NAME fix.

2002-09-06	<mschimek@users.sf.net>

	* options.c, options.h, main.c: Added -k option.
	* systems/mpeg1.c, systems/systems.c: Added -k option.

2002-09-05	<mschimek@users.sf.net>

	* audio/mp2.c: Preliminary audio underrun fix by Nikolai Zhubr.

2002-09-04	<mschimek@users.sf.net>

	* devices/v4l.c: Fixed pthread_join bug.

2002-08-31	<mschimek@users.sf.net>

	* options.c: Fixed -n h:m:s, worked for -m2 only.

2002-08-26	<mschimek@users.sf.net>

	* devices/v4l.c: Applied corrections by Nikolai Zhubr.

2002-08-21	<mschimek@users.sf.net>

	* video/mpeg1.c: Rewrote aspect ratio encoding.
	* video/mpeg2.c, main.c: Dipped a toe into MPEG-2 interlace.
	* video/filter.c, video/filter_mmx.c: Added YUYV #2 and #3,
	  horizontal and bidirectional 2:1.
	* devices/v4l.c: Applied read fix and AIW patch by Nikolai Zhubr.
	* options.c, options.h: New AIW hack, frame rate and aspect
	  override options.
	* video/mblock.c: Added video_sampling_aspect() helper.

2002-07-25	<mschimek@users.sf.net>

	* video/mpeg1.c: Corrected NTSC aspect ratio code.

2002-06-06	<mschimek@users.sf.net>

	* video/vlc_gen.c: Added. There's just no reason to create static
	  tables at runtime and it's cleaner too.
	* video, audio: Removed device interfaces.
	* devices: Moved device interfaces here, cleanup for rte.

2002-05-29	<mschimek@users.sf.net>

	* video/v4l.c: Applied patch by Monty Walls.

2002-05-27	<mschimek@users.sf.net>

	* video/mpeg1.c: Added rte_status.valid flags.

2002-05-13	<mschimek@users.sf.net>

	* video/filter.c: Fixed YUV 4:2:0 chroma offset bug.
	* Finished transition to new filter interface and added
	  faster SSE and YUV 4:2:0 versions.
	* global_data.c: Default now -F8 instead of -F3, trying
	  YUV 4:2:0 before 4:2:2.

2002-05-10	<mschimek@users.sf.net>

	* video/mpeg1.c, audio/mp2.c, options.c: Added -t0x200. 

2002-05-09	<mschimek@users.sf.net>

	* common, configure.in: Dereferenced symlinked videodev files,
	  was useful at one time but now just a nuisance.
	* video/mpeg1.c, systems/mpeg1.c: Added fifo usage test.
	* b_mp1e.c: Added split_sequence dummies.
 
2002-05-06	<mschimek@users.sf.net>

	* video/mpeg1.c: Fixed a recently introduced quantization
	  in intra pictures bug.

2002-05-04	<mschimek@users.sf.net>

	* audio/mp2.c, audio/audio.h: Changed num_frames option to double,
	  modified termination to permit seamless splitting.
	* video/mpeg1.c, video/video.h: Changed num_frames option to double.
	* systems/systems.c, systems/mpeg1.c, systems/systems.h: Extended
	  for seamless splitting. 
	* options.c, options.h, global_data.c: Changed frame
	  counters to double, added -z|--split option.
	* main.c: Changed frame counters to double, added functions to
	  reopen output file for -z|--split.
	* mp1e.1: Added -z, --split section.

2002-04-11	<mschimek@users.sf.net>

	* Added clock drift code switch (-t256).
	* video/mpeg1.c: Made coded_frame_rate option non-displayable.

2002-04-10	<mschimek@users.sf.net>

	* b_mp1e.c: Couldn't link rte w/o outFileFD and mux_syn.

2002-04-09	<mschimek@users.sf.net>

	* main.c: Virtual frame rate was limited to capture rate
	  instead of coded rate.

2002-03-22	automake 1.6 <mschimek@users.sf.net>

	* configure.in: Prepared for automake 1.6.
	* Automake.am: Removed global_data.c from libmp1e_la_SOURCES,
	  1.6 doesn't like it and it's no longer needed anyway.
	* b_mp1e.c: Added mux thread, seems to work.
	  More tests required.

2002-03-16	<mschimek@users.sf.net>

	* b_mp1e.c: Added mux i/o, untested.

2002-03-14	<mschimek@users.sf.net>

	* systems/vcd.c: VCDImager requires n * sector size file.
	* main.c: Added warnings about audio bit rate and vcd
	  parameter limits.

2002-03-11	<mschimek@users.sf.net>

	* options.c: Fixed typo in -o error check. -a, -m,
	  -X etc accepted out of bounds values.
	* audio/mp2.c: fixed option_get(num_frames) bug,
	  num_frames = INT_MAX ~ oo.
	* video/mpeg1.c: num_frames = INT_MAX ~ oo.

2002-02-18	<mschimek@users.sf.net>

	* mblock.c: Fixed video_look_ahead().

2002-02-08	<mschimek@users.sf.net>

	* Added -o patch by Xye
	* Added SIGUSR1 patch by Jon Hittner

2002-02-05	<mschimek@users.sf.net>

	* New option -n h:m:s record time.
	* Rewrote VCD mux.
	* b_mp1e.c: Changes towards rte 0.5.

2002-01-23	Fixes <mschimek@users.sf.net>

	* systems/systems.c: Fixed stupid bug in stream_sink().

2002-01-18	Fixes <mschimek@users.sf.net>

	* Improved timestamping in audio&video interfaces.
	* audio/mp2.c: Fixed audio frame dropping bug.
	* audio/arts.c: Added, ported from Zapping. (Just for
	  completeness, stinks no less than the ESD interface.)

2002-01-13	Fixes <mschimek@users.sf.net>

	* test/preview.c: Restored.
	* video/mpeg1.c: Fixed bug in frame skipping code. 

2001-12-11	<mschimek@users.sf.net>

	* video/mpeg1.c: executed MC preparation despite the picture not being
	  referenced. (However in normal IPB streams this occurs just once.)
	* b_mp1e.c: Fixed coded frame rate setup.

2001-12-05	<mschimek@users.sf.net>

	* Added pmmx_YUV420_0, pmmx_YUYV_0 (experimental).

2001-12-03	YUV stream interface <mschimek@users.sf.net>

	* Added, contributed by Harm van der Heijden.

2001-11-28	Multiplexer fixes <mschimek@users.sf.net>

	* systems/mpeg1.c, mpeg2.c, vcd.c: Fixed I picture PTS reorder
	  offset bug, created incorrect I PTS when encoding B frames
	  (not in 1.9.1).
	* -e0: Fixed incorrect P picture offset.
	* -e1: Fake B picture PTS not coded anymore, fixed incorrect
	  I/P PTS offset in video/mpeg1.c.
	* -e2: worked. :-)

2001-11-27	<mschimek@users.sf.net>

	* v4l.c: Try VIDEO_PALETTE_YUYV before _YUV422. (Some drivers
	  do not support _YUYV, and YUV422 may be YUYV (bttv), UYVY
	  (dc10) or according to v4l api spec YYYYUUVV.)
	* psycho.c: Fixed array initialization, segfault was possible.

2001-11-20	<mschimek@users.sf.net>

	* motion.c fix: ignored result 2nd of (0,0) test (rare, not critical).
	* mp1e.1: Corrected -F typo in examples.
	* video/mpeg1.c: Frame skipping flaws. Rewrote.

2001-11-06	<mschimek@users.sf.net>

	* configure.in: Corrected $AS.
	* v4l.c: grab size / capture size / decimating fixed.
	* v4l2.c: 288->240 fallback fixed.

2001-11-01	<mschimek@users.sf.net>

	* mp2.c: Fixed audio table selection (wrong sampling_freq).
	* oss.c: Mixer always set volume of line, not requested source.
	* main.c: Added fallback for unsupported joint stereo.
	* mp1e.1: Corrected a few typos.

2001-10-23	<mschimek@users.sf.net>

	* mp2.c: Improved the codec status stuff, moved the sample
	  parameters code.
	* systems.c: Made buffer allocation in add_input_stream
	  optional to clean up init() in mp2.c.
	* b_mp1e.c: Added format enumeration code, unfinished.

2001-10-21	<mschimek@users.sf.net>

	* video/mpeg1.c: Fixed incorrect type of bit_rate, strdup(0)
	  in option_print, virtual > coded_frame_rate bug.
	* audio/mp2.c: mpeg1_mp2_thread -> class.mainloop. 

2001-10-18	<mschimek@users.sf.net>
	
	* esd.c: Replaced timestamping code by the v4l type, should
	  better handle audio dropping.
	* oss.c: Accept other formats than S16LE.

2001-10-17	<mschimek@users.sf.net>

	* oss.c: Fixed sampling_freq check, reported by Ralf Oehler.
	* alsa.c: Corrected 0.9 compiling errors.

2001-10-14	<mschimek@users.sf.net>

	* mpeg1.c: Fixed drop & stop bug, fixed off-by-one frame_rate_code.
	* added rte_parameters and status.

2001-10-01	<mschimek@users.sf.net>

	* audio: Polished the timestamping, some bad changes since
	  09-20. Still not what I'd like, but hopefully sufficient.
	* mpeg1.c: Rewrote frame dropping/skipping code, added rte
	  options, lots of small changes.

2001-09-28	<mschimek@users.sf.net>

	* v4l2.c: Added Steven Davies' bttv 0.8 v4l2 r/w patch.

2001-09-25	<mschimek@users.sf.net>

	* video: Some work towards reentrancy and encapsulation.
	* mp2.c: Added S|U8|16LE sample conversion.
	* alsa.c: Added new direct, mmapping interface. Let's see
	  if this works better.

2001-09-23	finished rte audio options <mschimek@users.sf.net>

	* Moved rte_option stuff from b_mp1e.c to mp2.c, added get_option.
	* Made audio reentrant, unlimited number of mp2 streams, prefixed
	  local functions, all tests compiled in.
	* Changed mp1e frontend to use rte codec interface for audio.

2001-09-20	audio clock drift extravaganza <mschimek@users.sf.net>

	* oss.c: Found a timestamp bug (ispace sign, code obsolete), audio
	  data wrap-around code removed, timestamping rewritten from scratch,
	  explict allocation of DMA fragments, OSS optional.
	* alsa.c: Replaced audio data wrap-around by flat buffer code, added
	  timestamp low-pass, added ALSA-0.9 interface (preliminary, is
	  untested and needs better time stamping), both optional.
	* esd.c: Replaced audio data wrap-around by flat buffer code, added
	  timestamp low-pass (more a protection for the clock drift algo,
	  there's no good method to avoid a constant lag caused by esd
	  overhead, if any).
	* afl.c: Replaced audio data wrap-around by flat buffer code,
	  fixed sampling rate bug introduced 2001-07-27, AFL wasn't
	  mentioned in the man page although available in earlier versions.
	* mp2.c: Bug in audio sync code (obsolete), rewrote sampling
	  device interface from scratch, removed zero-copy option, added
	  clock drift compensation code.
	* sync.c: Extended sync_break for clock drift compensation.
	* configure.in: New alsalib test, driver interface summary.
	* v4l.c: Added timestamp low-pass (esd.c comment applies).

2001-09-13	user data to mux_alloc <garetxe@users.sf.net>

	* systems: pass down user_data to mux_output, needed for getting
	rid of rte_global_context.

2001-09-12	<mschimek@users.sf.net>

	* Made the initial audio sync sample accurate (admit that's a bit
	  overkill) and fixed a minor bug in sync.c.

2001-09-07	audio noise bug <mschimek@users.sf.net>

	* afl.c, alsa.c, esd.c: Fixed an internal interface bug.
	* global_data.c: PCM device didn't default to ALSA if present.
	* options.c: Cleaned up usage text.

2001-09-03	audio sync hang fix <mschimek@users.sf.net>

	* sync.c: Added bytes_per_sample parameter, for (;;) aborts
	  after two seconds.
	* audio.c: Pass source, not dest frame rate to sync(),
	  follow-up to relaxing the input buffer size.

2001-09-01	<mschimek@users.sf.net>

	* configure.in: Added create_links(), debugging option, mmxemu test.
	* main.c: Uses mmxemu if available for debugging.

2001-08-29	<mschimek@users.sf.net>

	* video/filter.c: init_filter looked suspicious, rewrote.
	* b_mp1e.c: Added image size error check.
	* mpeg1.c: Fixed "dnl gcc3 -O0 makes mmx_mbsum segv, why?". Stupid me.

2001-08-27	post-release fixes <mschimek@users.sf.net>

	* configure.in: Temporary change of the alsa test due to 0.9
	  incompatibility.
	* global_data.c: Back to default -F1, was a bad idea apparently.

2001-08-19	rte fix <mschimek@users.sf.net>

	* mp1e_b.c didn't select a multiplexer, static default mpeg1-ps,
	  set to elementary when only one stream present for mp1e cli compat.
	* Improved checks in multiplexers.
	* Cloned AC_CHECK_FUNCS(program_invocation_name) from Zapping.

2001-08-10	<mschimek@users.sf.net>

	* Managed to confuse Intel and AT&T syntax in motion.c only
	  newer 'as' would understand. Converted to AT&T.
	* video/v4l.c: Didn't work, rewrote, let's see.

2001-08-08	<mschimek@users.sf.net>

	* New mux context. (Wanted to replace output stuff but that
	  needs more preparation.)

2001-08-06	<mschimek@users.sf.net>

	* ffmpeg: crash due to cpuid bug, updated from ffmpeg cvs.
	* rte/b_mp1e: report audio frame count in audio only mode.
	* b_mp1e: increased audio buffer size to 8 * frame size (similar
	  to mp1e btw.), hope this gives better sync w/esd. Data is more
	  likely new when we read/stamp because encoding is much faster
	  than input, although still far from reliable synching.
	* rte: fixed a typo in destroy sequence causing hangs.
	* mp1e video: missing eof reply.

2001-08-01	<garetxe@users.sf.net>

	* ffmpeg/: Added ffmpeg backend, taken from ffmpeg cvs
	(http://ffmpeg.sf.net). Needs a lot of work, but most modes
	already encode.

2001-08-01	<mschimek@users.sf.net>

	* video/mpeg1.c: Fixed fifo eof handling.
	* Restored preview, preliminary.
	* Examined the 2001-06-07 bug, is a DC rounding problem,
	  not strictly a bug. Modified the quant calculation.
	  Lets see how that works, then optimize.
	* Restored v4l interface.

2001-07-31	<mschimek@users.sf.net>

	* Added optional input fifo wrapper to mp2 encoder.

2001-07-28	<mschimek@users.sf.net>

	* Converted remaining fifos to fifo2.

2001-07-27	<mschimek@users.sf.net>

	* Converted capture fifos to fifo2 in mp1e and rte.
	* Configure checks whether as supports SSE2.

2001-07-18	<mschimek@users.sf.net>

	* Rewrote bstream function
	* Several fixes and improvements in emu
	* Added SSE2 support

2001-07-13	<mschimek@users.sf.net>

	* Fixed video API detection
	* Oops, forgot to clean up SIGTERM

2001-07-12	<mschimek@users.sf.net>

	* Made emu work with pthreads (phew...)
	* Added SSE and 3DNow! support

2001-07-06	<mschimek@users.sf.net>

	* Made ioctls EINTR aware
	* Corrected a B quant typo (post 1.9pre1)
	* Added motion option to rte

2001-07-05	<mschimek@users.sf.net>

	* Fixed global vbuf conflict in rte_test, same correction
	  in video/v4l2.c, just in case

2001-07-02	<mschimek@users.sf.net>

	* Added man page

2001-06-22	<mschimek@users.sf.net>

	- MC + 704x576 -> segv
	  2001-07-05 can't reproduce
	* gcc 3.0 fixes

2001-06-07	<mschimek@users.sf.net>

	* Restored subtitle recording
	- MC P/B bug still with us :-P

2001-06-05	<mschimek@users.sf.net>

	* Removed all hardcoded image sizes from motion code
	* Some cleanup in motion.c
	* Integration of motion finished (not enabled by default)

2001-06-04	<mschimek@users.sf.net>

	* Fixed B-frame VLC bug
	* Fixed B-motion mb skipping bug
	* Fixed B-motion PMV bug
	* Fixed MV overflow bug (could be improved)

2001-06-01	<mschimek@users.sf.net>

	* Rewrote configure.in; added config.h
	* Fixed closed_gop B-motion bug
	* Integrating motion, unfinished

2001-05-27	<garetxe@users.sf.net>

	* Fixed rte buffered push capture, removed the possibility of
	deadlock in kill_zombies (the solution isn't optimal,
	though). rte_test works now with all possible combinations for me.
	* YVU support.

2001-05-10	<garetxe@users.sf.net>

	* Added .spec.in file, mux_thread_done flag for cleaner exit, some
	more fixes to rte/fifos. Lots of useless debugging of mpeg plugin
	segfaults/hanging, cannot isolate the bug yet.

2001-05-07	<mschimek@users.sf.net>

	* Added MC 3.1, not yet compiled in by default, needs testing.
	- MC: stream syntax bug, probably large vectors
	  [-> confirmed, fixed on 2001-06-04]
	- With MC enabled: P/B block bug (quant?), apparently at scene cuts

2001-05-06	<garetxe@users.sf.net>

	* more work on fifos (killing zombies, fixes). Still some bugs
	left.
	* (later) fixed a couple of bugs, plus a purely theoretical one.

2001-04-12	<mschimek@users.sf.net>

	* Fixed p-frame skipped mb bug.