changelog

mpeg 2 source code for video compression

	* TODO: note about track gain adjustment

2006-07-24 01:05  thor

	* BENCHMARKING.thor, mpg123.1: in BENCHMARKING now

2006-07-24 01:04  thor

	* BENCHMARKING: Updated the benchmarking notes, added
	  BENCHMARKING.thor contents as well as some fresh recheck.

2006-07-24 00:56  thor

	* INSTALL: updated installation instructions

2006-07-24 00:15  thor

	* src/mpg123.c: fixed lecagy build broken by VERSION_STRING; made
	  help output nicer

2006-07-23 22:58  thor

	* src/Makefile.legacy: fixing the install target

2006-07-23 18:56  thor

	* AUTHORS: cleaning up and warning spammers

2006-07-17 22:37  thor

	* mpg123.1, src/mpg123.c: - seen that manpage needs a lot of
	  updating (nearly ten years old!) - made the wanted usage info
	  (-?, --help, --longhelp) go to stdout and only the help-on-error
	  to stderr - added --version to give program version string

2006-07-17 21:04  thor

	* tools, tools: keep the main trunk tight; auxilliary tools in
	  separate location

2006-07-17 21:03  thor

	* precompiled: we have assemblers nowadays, no need to keep the
	  precompiled objects

2006-07-17 19:45  thor

	* BUGS, CHANGES, CONTACT, LICENSE, README, README.cfa, README.new,
	  README.remote, README.thor, src/control_generic.c: Getting
	  documentation closer to release, typo in control_generic help.

2006-07-17 19:25  thor

	* README: to be replaced by new one

2006-07-17 17:33  thor

	* legal/statements/Nicholas_J._Humfrey.email, COPYING: Updated
	  COPYING to a near-final state, added agreement hint by Nicholas.

2006-07-17 16:20  thor

	* legal/first_final.txt, src/Makefile.am, src/layer1.c,
	  src/layer2.c, src/layer3.c, src/layer3.h, src/mpg123.c,
	  src/mpg123.h, src/readers.c, src/system.c, src/tabinit.c,
	  src/term.c, src/term.h, src/wav.c, src/xfermem.c, src/xfermem.h:
	  Next round of boilerplates along with the story. With the
	  current argument, there is only one GPL-only file left!

2006-07-17 16:16  thor

	* MakeLegacy.sh: We write C, not C++.

2006-07-17 14:59  thor

	* AUTHORS, COPYING, src/audio_esd.c, src/equalizer.c, src/genre.h,
	  src/getbits.c, src/getbits.h, src/getcpuflags.s, src/getlopt.c,
	  src/getlopt.h, src/httpget.c, src/huffman.h, src/l2tables.h:
	  equalizer patch really was jsut a path to equalizer, new author
	  for linux-ppc-nas make target

2006-07-17 14:57  thor

	* src/Makefile.legacy: added PPC/NAS target from
	  <[e-mail]>

2006-07-17 13:54  thor

	* src/dct64_MMX.s, src/decode_MMX.s, src/tabinit_MMX.s: the MMX
	  stuff came into Michaels hands without comment -> LGPL by author
	  rule

2006-07-17 13:53  thor

	* AUTHORS: credit for Mikko and his DCT9

2006-07-17 03:22  thor

	* lgpl/first_final.txt, AUTHORS, COPYING, scripts/debugdef.pl,
	  src/Makefile.legacy, src/audio.c, src/audio.h, src/audio_aix.c,
	  src/audio_alib.c, src/audio_dummy.c, src/audio_esd.c,
	  src/audio_hp.c, src/audio_jack.c, src/audio_libao.c,
	  src/audio_macosx.c, src/audio_mint.c, src/audio_nas.c,
	  src/audio_os2.c, src/audio_oss.c, src/audio_sgi.c,
	  src/audio_sun.c, src/audio_win32.c, src/buffer.c, src/buffer.h,
	  src/common.c, src/common.h, src/control_generic.c,
	  src/dct36_3dnow.s, src/dct64.c, src/dct64_3dnow.s,
	  src/dct64_MMX.s, src/dct64_altivec.c, src/dct64_i386.c,
	  src/dct64_i486.c, src/debug.h, src/decode.c, src/decode_2to1.c,
	  src/decode_3dnow.s, src/decode_4to1.c, src/decode_MMX.s,
	  src/decode_i386.c, src/decode_i486.c, src/decode_i586.s,
	  src/decode_ntom.c, src/equalizer_3dnow.s, src/tabinit_MMX.s:
	  First part of boilerplating and sorting all out. Looks kinda
	  nice now, still would like to get the higway mmx stuff clear...
	  maybe Michael remembers something there.

2006-07-17 02:15  thor

	* AUTHORS: little update for MMX, unnecessary verbosity about what
	  maintaing implies

2006-07-17 01:30  thor

	* src/audio_jack.c: boilerplate and // -> /* */ as mpg123 is
	  (still) an old-compiler friendly project

2006-07-17 01:02  thor

	* src/common.c, src/control_generic.c, src/httpget.c,
	  src/layer3.c: Oh, I should compile more often with strict
	  settings: pedantic gcc complained about some C89 violations I
	  introduced. Fixed that.

2006-07-17 01:01  thor

	* configure.ac, src/audio_alsa.c, src/audio_alsa09.c: Old alsa
	  output doesn't work with any sane setup using alsa today
	  (meaning: not using the 0.5 api). New doesn't work because
	  unfinished yet. Alsa output actually is available through libao.
	  So, deactivated alsa stuff in configure for now...

2006-07-17 00:06  thor

	* Makefile.am, autogen.sh, configure.ac: more boilerplates

2006-07-16 23:56  thor

	* MakeLegacy.sh: boiler

2006-07-16 23:54  thor

	* scripts/debugdef.pl: public domain should be gpl compatible...

2006-07-16 23:46  thor

	* mpglib/trunk, mpglib: mpglib will be something of its own, taken
	  care of later...

2006-07-16 23:39  thor

	* Makefile.win32, test.pl: niceness by a line

2006-07-16 23:38  thor

	* test.pl: boilerplate, style cleanup (could use more use, though)

2006-07-16 23:37  thor

	* Makefile.win32: boilerplate

2006-07-16 23:24  thor

	* src/audio_oss.c: just a paranoid note

2006-07-10 02:23  thor

	* src/audio_libao.c: Polished the libao output a bit while testing
	  it on my linux laptop: - parsing -a driver:file for file devices
	  (well, this could be extended to a general scheme to choose
	  between different driver subsystems via -a system:file or
	  system:device:file ...) - corrected the error handling: -1 is
	  expected on error by mpg123.c - giving help message with list of
	  available drivers when invalid one is specified I verified with
	  one testfile that -a wav:file does the same as -w file. I think
	  that this is good;-) Also, I'd like to mention that it's nice to
	  measure audio subsystem overhead with comparing -t and -a null.
	  With the libao in place, I'm wondering if we still want our own
	  alsa output... let's see...

2006-07-09 17:37  thor

	* buglog/fuzzing.txt, buglog/mathias_wenzel_stress.txt,
	  src/layer3.c: next segfault by fuzzing... fixed this special
	  test case but one needs full understanding of layer3.c to really
	  get it

2006-07-06 14:00  thor

	* buglog/mathias_wenzel_stress.txt, src/common.c, src/layer3.c:
	  Fixed the mpeg2.5 verdict in common.c (we have to check support
	  for this); made parts of layer3.c more robust based on an evil
	  mp3 (see the buglog file). Most important featur of this
	  robustness could be that mpg123 doesn't simply exit() in wrong
	  sideinfo anymore but tries to continue.

2006-07-05 20:45  thor

	* TODO, src/httpget.c: Revised my bugfix for the last GLSA issue
	  after suggestion by Daniel Kobras (allow longer redirection urls
	  than request url), spent some time hacking further ... and: -
	  fixed unnoticed memory leak in redirection (malloc without free
	  for assigned pointer) - now accepting non-absolute redirects;
	  request now is always of the form GET http://host/path HTTP/1.0
	  (non-absolute relocations are related to this request) That also
	  works flawlessly with my virtual domain setup on my server used
	  in the test. - decided that this whole file is evil; we need to
	  rewrite that properly

2006-07-04 12:41  thor

	* buglog, buglog/buglog.txt,
	  buglog/gentoo-bug133988_GLSA_200607-01.html,
	  tags/0.59r-thor6/mpg123.c, src/httpget.c: Reacted to gentoo bug
	  133988 / GLSA 200607-01 and fixed it properly, storing and using
	  the allocated length of purl as well as ensuring string
	  termination.

2006-07-04 01:43  thor

	* src/common.c, src/layer3.c, src/layer3.h, src/mpg123.c: Again
	  reworked the gapless stuff... should be more efficient now and
	  also more aggressive by just skipping decoder delay even if no
	  info found and also adding decoder delay to end (hm, where does
	  this really make sense without padding removal?). It should be
	  more efficient now; a comparison between linux-x86 binary
	  (pentium) with and without gapless code shows perhaps around
	  0.2% speed penalty through added code - which already melts when
	  gapless mode is actually active. Generally on speed: comparison
	  with 0.59r (linux build, non-marched cflags (pentium2)) shows
	  around 2% slower decoding for svn trunk; but this is greatly
	  compensated by using one of the new assembler optimizations.
	  Still, I see much code that can be improved also for efficiency
	  alone by clearing the situation with some (shared) variables
	  that don't need to be passed around all the time...

2006-07-04 01:25  thor

	* src/Makefile.legacy: these -g were leftovers

2006-07-03 12:55  thor

	* src/Makefile.legacy, src/common.c, src/common.h,
	  src/control_generic.c, src/layer3.c, src/layer3.h, src/mpg123.c,
	  src/mpg123.h: Main message: The gapless mode has been enhanced
	  to work more low-level (in layer3.c) and also to use the
	  Xing-given frame count to get the end straight. It should work
	  (tested, but not in every detail) both with the -k option and
	  terminal controlled seeking as well as control_generic JUMPs.
	  The current implementation adds some load that could be reduced
	  by putting the gapless checker code into behind flush check
	  instead of before but that needs additional care for track end
	  (where audio_flush is called externally). Apart from that and
	  some needed de-uglification it should be a nicely working
	  feature. Minor messages: frameNum is unsigned long (so bigger
	  that int and never < 0!); some parts of mpg123.c and common.c
	  did not honor these truths. But I don't claim that I catched all
	  occurences of such problems - see BUGS. Also, I fixed stuff in
	  print_stat with weird frame number estimations. Looks _much_
	  better now esp. when the frame count is known from Xing tag.

2006-07-03 12:16  njh

	* AUTHORS: Added Romain Dolbeau to AUTHORS file for his Altivec
	  work

2006-07-03 12:12  njh

	* configure.ac, src/Makefile.am, src/dct64_altivec.c: Imported
	  altivec support from mplayer (and it works!) Still need to do
	  CPU detection.

2006-07-03 12:11  thor

	* TODO: more and less of this

2006-06-27 00:54  thor

	* src/config.h.legacy: Did we start using C++ style comments?

2006-06-27 00:51  thor

	* src/common.c: a note on gapless hack updated

2006-06-26 22:57  thor

	* README.new: added note about more natural Ctrl+C

2006-06-26 22:43  thor

	* src/Makefile.legacy: some more flags by default

2006-06-26 22:40  thor

	* src/Makefile.legacy: ' seems to be dangerous together with
	  line-continuing \ ... made more "

2006-06-26 22:22  thor

	* configure.ac: generic linux support (?)

2006-06-26 22:04  thor

	* src/mpg123.c: Following DK's suggestion in a comment, I disabled
	  the Ctrl+C (SIGINT) hack when a control mode is active. You can
	  still do Ctrl+C to skip a song and twice to end playback if you
	  do not use the terminal or remote control mode. With active
	  control mode, a Ctrl+C now kills as expected (formerly, it was
	  just ignored).

2006-06-26 21:36  thor

	* src/getlopt.c, src/mpg123.c: There were several option parsing
	  regressions most probably caused by application of Nicholas'
	  changes to my code with changed command line parsing. Fixed the
	  obvious ones and made the option parser bail out if it encouters
	  such an error again.

2006-06-26 21:34  thor

	* MakeLegacy.sh, src/config.h.legacy: More legacy make system
	  working again with more crafted config.h.

2006-06-26 13:17  thor

	* AUTHORS, README.new: history in README.new, maintainers in
	  AUTHORS

2006-06-26 12:24  thor

	* src/audio.c, src/audio_aix.c, src/audio_alib.c,
	  src/audio_alsa.c, src/audio_dummy.c, src/audio_esd.c,
	  src/audio_hp.c, src/audio_jack.c, src/audio_libao.c,
	  src/audio_macosx.c, src/audio_mint.c, src/audio_nas.c,
	  src/audio_os2.c, src/audio_sgi.c, src/audio_win32.c,
	  src/buffer.c, src/common.c, src/control_generic.c, src/dct64.c,
	  src/dct64_i386.c, src/dct64_i486.c, src/decode.c,
	  src/decode_2to1.c, src/decode_4to1.c, src/decode_i386.c,
	  src/decode_i486.c, src/decode_ntom.c, src/equalizer.c,
	  src/getbits.c, src/httpget.c, src/layer1.c, src/layer2.c,
	  src/layer3.c, src/readers.c, src/system.c, src/tabinit.c,
	  src/wav.c, src/xfermem.c: Added inclusion config.h everywhere to
	  make sure we don't stumble over some missing define later and
	  search like crazy...

2006-06-26 12:22  thor

	* src/mpg123.c: Sorry, but these pointlessly broken lines were
	  _ugly_.

2006-06-26 12:11  thor

	* src/getlopt.c, src/term.c: These got some new-style debugging
	  through the terminal control bug hunt. Also, I want to take the
	  chance to hint at scripts/debugdef.pl that I added previously:
	  It is a Perl script generating the debug.h header in a way to
	  avoid variadic macros for fprintf. Result are debugN macros that
	  mean "debug line with N arguments". Examples: debug1("Ok,
	  there's one value: %i", var) debug3("now three of them: %i %f
	  %i", var1, var2, var3) simplest case: debug("just a message");

2006-06-26 12:06  thor

	* scripts, scripts/debugdef.pl, src/debug.h, src/mpg123.c: Hah!
	  Found the place where terminal control got killed! The option
	  definition in mpg123.c lacked the type specifier (a regression
	  caused by merging). To find that, I added debug.h for more
	  convenient debugs.

2006-06-26 01:05  thor