changes

mpeg层3音频编码程序

###---> in 0.7.1
- indicate that we don't know how to decode MPEG2.5. sorry.
- a dummy crc check (we can play crc protected files now)
- an altogether better huffman decoding - a complete rewrite
- skip first, or first few, frames if main_data_begin is non zero/too big
- fixed mono playing - a bug in transform.c
- fixed /dev/dsp initialization 
- included Karl's IRIX port
- fixed a bug in fras_l and fras_s
- added a part of huffman table 13 that was missing
- included Karl's optimization patch - a huge speedup
- included Lutz' HP/UX port

###---> in 0.7.2
- added clipping code
- fixed a bug in huffman decoding
- included Dan Nelson's FreeBSD port
- Edouard's xmpeg3 gui is a part of the distribution now

###---> in 0.7.3
Andrew Richards:
- Reorganised audio.c (alot!)
- Moved the audio device specific stuff to separate architecture files
- Added a pipe based audio buffer
- Added an args.c for argument processing with getopt_long()
- Added preliminary volume control
- Added the ability to play more than one file
- Added stdin and stdout support
- Added configure support
- Added some debugging features
Karl Oygard:
- optimised the Huffman decoder 
- more optimisations in transform.c, misc2.c, getbits.c, ...
Tomislav Uzelac:
- new imdct algorithm (thanks to Jeff Tsay, Mikko Tommila)
- minor bugfixes 

###---> in 0.7.4
- moved the initial memcpy() from audio.c into fillbfr() itself
- enhanced synchronization code 
- indicate that we don't know how to decode layer 1 or 2 (yet)
- indicate that we don't know how to decode free format bitstreams
- more optimisations (by Karl, of course), but no assembly yet 
- IRIX support fixed (Karl)
- AIX code included (George Colouris)
- -fomit-frame-pointer option reinstated
- ffwd/rew capabilities added
- buffer implementation improved (Karl)
- sajber jukebox support in place (Edouard)
- layer2 is supported (not finished though)
- PAUSE bug fixed (Lutz)
- BeOS port integrated into the main source tree (Andy Lo A Foe)
- Linux realtime sound driver support (Thomas Sailer)
- most of poly() re-written in assembly for i586 (Karl)
- speedups in requantize_ms
- a new configure script
/* 0.7.4 never actually got released ... */

###---> in 0.7.5-0
- fixed the core dump when using rt with -buffer option
- removed buffer status display
- show_header rewritten

###---> in 0.7.5-1
- fixed intensity stereo bug
- moved gui control stuff to guicontrol.c
- removed a call to AudioBufferFlush() when gui is paused

###---> in 0.7.5-99
- fixed Kyrandia_Forrest bug (in imdct)
- downmix stereo->mono
- fixed mono playing (Justin Frankel)
- changes to configure

###---> in 0.7.6
- added initialise_decoder(); t_43[] is now initialised on startup
- make amp just die in case euid!=0 when rt is enabled (temporary)
- disabled -convert with rt since it doesn't work (tmp)
- removed ld()
- added a generic get_input() to switch between prefetch/plain reads
- created initialize_globals()
- added a call to initialize_globals() to decoder_process()
- fixed premature end of playing in rt
- fixed exit condition in prefetch_get_input(); short files are now OK
- removed snd_eof from globals (not needed)
- commented out perror()s in rt_play (unneeded error messages sometimes)
- fixed IRIX premature end of playing (Karl)
- fixed a bug in gethdr() which affected recognition of MPEG2.5 files
- added xxx_fancy_audio() functions to encapsulate direct-to-DMA stuff
- reorganized audio.c to enable merging
- re-fixed a problem with some very old mp3 files (huffman tables 4 & 14)
- /dev/dsp is now opened in setup_fancy_audio
- commented out cinfo.ptr=0 from start_fancy_audio()
- more new functions: prefetch_initial_fill(), set_realtime_priority()
- realtime patches for HPUX (Lutz) - NOT FINISHED YET!!!
- merged realtime and non-realtime decoder loop (yeah!!!)
- resetting s and res to 0 in initialise globals. (Leif, let me know
  if this helps)
- configure changes, mainly for SunOS and HPUX