view CHANGELOG @ 172:705dcbf94639

Sound_DecodeAll() is now more robust.
author Ryan C. Gordon <icculus@icculus.org>
date Sat, 01 Dec 2001 01:12:58 +0000
parents 697fabe1453e
children 865c17682d8f
line wrap: on
line source

/*
 * CHANGELOG.
 */

11302001 - Torbjörn and I make Sound_DecodeAll() more robust: checks for
           previous decoding failures and sets an appropriate error, handles
           decoders that change their buffers on the fly (such as the FLAC
           decoder), and deals with out-of-memory conditions more gracefully.
11252001 - (With thanks to Andreas Umbach for pointing it out) Fixed some
           problems with Sound_DecodeAll(). For local testing of this bug,
           added a --predecode command line to playsound. Minor fixes to
           theoretical bugs in Sound_FreeSample(). playsound no longer
           buffers stdout and stderr. Updated Sound_DecodeAll()'s comments in
           SDL_sound.h ...
11192001 - FLAC decoder cleanups from Torbjörn.
11092001 - Torbjörn fixes playsound's audio callback after I broke it, again.
           A bug in configure.in was preventing SMPEG from being used unless
           --enable-debug was set; fixed. Changed this file to list latest
           changes first. Torbjörn submitted a FLAC decoder that utilizes
           libFLAC (http://flac.sf.net/). Cool.
11012001 - API COMPATIBILITY BREAKAGE: Decoders can now list multiple file
           extensions each. Playsound has been updated to handle this.
           Playsound now registers a SIGINT handler, so you can skip tracks
           and/or abort the way that mpg123 does.
10232001 - Rewrote playsound.c's audio_callback() to no longer need the
           overflow buffer hack, which streamlines it a little and trims the
           memory requirements for playsound by about 16 kilobytes.
10172001 - Torbjörn catches a problem with the overflow buffer in playsound's
           audio callback.
10152001 - Torbjörn sends in a default sample format for the MIDI decoder, 
           and the starts of the audio conversion funcitonality (ripped 
           from SDL). Officially released 0.1.3. Added LICENSE and 
           CHANGELOG to the distribution. (Again, from Torbjörn) added in 
           the start of a tweaked audio converter.
10122001 - Torbjörn Andersson submitted command line enhancements to 
           playsound, and I cleaned up the --help output.
10092001 - Patches to shn.c for Visual C compatibility. Visual C project files
           available from the website. Changed Corona688 to Tyler Montbriand
           in CREDITS. Upped version to 0.1.3.
10082001 - Restructured decoders/wav.c to allow for multiple formats, and 
           put the start of a handler for the ADPCM format in place.
10072001 - Changed the way decoders/mod.c handles samplerate so that it should
           work universally. This isn't an idea solution, but it's probably
           the best we can do without rewriting mikmod. Made a change to ogg.c
           for portability: changed an int64_t to ogg_int64_t.
10062001 - Made a change to SDL_sound.c for compiling on non-GNU toolchains.
10052001 - Removed #include "SDL_endian" from aiff.c.
10042001 - Changed some #if (defined SOUND_SUPPORTS_*) lines to
           #ifdef SOUND_SUPPORTS_* in voc.c and shn.c, for consistency with
           the other decoders.
10032001 - After hours of tracking down a bogus pointer, the SHN decoder works!
           I can die happy. :)  Max placated me with an --enable-debug option
           so I could stop my whining. Other autoconf goodies (such as
           reenabling -Werror for debug builds, etc). Torbjörn brings in a
           MIDI decoder, which reads from a Timidity process through a pipe.
           Changed playsound to open the audio device to match the properties
           of each sound file, which results in less conversion (and therefore,
           more chance of correct playback).
10022001 - Changed a comment in mod.c to not refer to "the mikmod 
           directory" anymore. Committed Torbjörn's patch for MP3 detection.
           (better late than never). __Sound_strcasecmp() now handles NULL
           strings gracefully, fixing the crash with "playsound bootstrap".
           More work on the SHN decoder.
10012001 - Fixed a memory leak that Torbjörn found in the MOD decoder.
09252001 - More autoconf work. Gave Max Horn write access to the CVS
           repository, so I don't drive him nuts tweaking this thing.  :)
           Fixed a const complaint and some other stuff needed for compilation
           under Visual C++ 6.0 (no, it isn't ported yet). Put the SHN source
           in CVS, even though it isn't ready (and doesn't even compile). Do
           NOT enable it in your build!
09242001 - Thank goodness, Torbjörn came through with the MP3 fix. Apparently
           SMPEG mixes each chunk of decoded data with whatever is already
           in the buffer you give it. I hate that. I'm going to patch SMPEG
           to let the programmer enable and disable that behaviour in a given
           (SMPEG *), since it's just a CPU eater in this case. The _D(())
           macro is now SNDDBG(()), since _D is taken on MacOS X's version of
           gcc (which was bound to happen on some platform sooner than later
           anyhow). Renamed test_sdlsound to playsound, and made it more
           robust in general: fixed potential overflow in audio_callback,
           made it chatter less, made it take multiple files and some other
           command lines. Initial autoconf support, thanks to Max Horn.
09222001 - Torbjörn Andersson strikes again, with a collection of patches.
           First, some cosmetic tweaks for decoders/aiff.c. Next, a MOD player
           based on MikMod. This inspired me to add two more methods to
           Sound_DecoderFunctions: init() and quit(). Third, a fix to
           decoders/mp3.c so that SMPEG won't claim every stream it sees, MP3
           or not. I removed the multiple-streams-per-rwops code, after
           discussion on the mailing list. The init() and quit() methods
           led to the possibility that certain decoders will flag themselves
           as unavailable at runtime, and SDL_sound now handles this.
           Added [LIB|INC]PATH_[OGG|MOD]. Bigendian fixes; now works on
           PowerPC Linux. MikMod tweaks. Changed version to 0.1.2.
09202001 - Torbjörn Andersson submitted several patches: fixed a comment in
           the .WAV decoder (whoops...screwed up my own search-and-replace.
           Hah.), made an attempt at putting multiple sound streams behind
           one RWops (gotta think on that one first), and, most importantly,
           added an AIFF decoder, which is very cool.
09192001 - Added a skeleton decoder source file. Changed voc_read() to
           voc_read_waveform(), so it wouldn't be confused with VOC_read().
           Fixed a byte ordering bug in voc.c (reported as AUDIO_S16LSB, but
           we were swapping byte order of data ourselves. Fixed). Added basic
           .WAV support. Fixed Makefile so that -I. is always first;
           otherwise, a previously installed header might get used for the
           compiles, which is not good. SDL_sound.h now includes SDL_endian.h,
           since SDL.h doesn't, for some reason. Moved version defines in
           SDL_sound.h to top of file so I can find them.  :)
           Changed version to 0.1.1. Committed patch from Tsuyoshi Iguchi to
           fix a segfault (I forgot to put a NULL terminator at the end of
           the available_decoders array), fixing the only bug preventing the
           test program from running on FreeBSD 4.3. Sweet. Added Ogg Vorbis
           decoder. Rewrote the test program's SDL audio callback to be more
           robust (Ogg exposed a nasty bug in it). Fixed a byte-ordering issue
           in the VOC decoder.
09182001 - Implemented MP3 support through SMPEG (not working yet, though) and
           wrote the Reference Counting RWops wrapper. Added other little
           things like the _D(()) macro. Added VOC support, which went up with
           surprisingly little struggle, which means it MUST be leaking
           memory.  :)
09172001 - Changed some overlooked "voice" to "sound". Implemented base API.
           So...tired. Everything's different.  :)
           Also put in a RAW decoder and a simple test program.
09142001 - Changed name to SDL_sound, added Sound_DecodeAll() to spec.
09132001 - Initial spec proposed on SDL mailing list, under name "SDL_voice".

--ryan. (icculus@clutteredmind.org)

/* end of CHANGELOG ... */