Latest changes to the source
============================


1999-08-07	Ok, back from the vacation I put 085-unstable online. 
		Improvements since 0.84-unstable:
		- Matthias Wächter found a quality degrading bug in his
		  optimisations for 0.84 and fixed it. He has also cleaned up
		  and optimized l3psy.c and subs.c a bit further, which should
		  do a few more percent.
		- Brad submitted a patch to make it compile nicely on OpenBSD.
		- Serg 'Ice' Tsyganenko found a bug in the code for reading the
		  commandline, which under certain circumstances could confuse
		  BladeEnc if one or more of the samples had ".mp3" in their body
		  or specified path.

		I have some more patches to merge, but I'm going to England for
		a long business trip now (possibly two months), so they will have
		to wait until I come back... :(

1999-07-30	Put 084-unstable online. Improvements since 0.83-unstable:
		- Received and added some really heavy optimisations made by 
		  Matthias Wächter, resulting in a 28% speed increase on my 
		  machine! 8-) These optimisations have however broken the
		  binary identicality with older versions (slight changes in 
		  precision (to the better) for a few routines early in the 
		  process) and I can therefore not guarantee that no quality
		  affecting bug has slipped through. The quick investigations
		  I've made of the code and output suggests that nothing bad 
		  has occured, but I would appreciate if some of you took the
		  time to test it more closely (frequency analysis and listening
		  tests mostly).
		- Fixed broken last frame in MP3-file using a sollution I know
		  isn't the right one, but as far as I can see it should do the
		  trick without any side effects. Thanks to Robert Schlabbach
		  for bringing it to my attention over and over again... ;-)
		- Fixed what seemed to be a quality affecting bug, but later
		  turned out to be harmless due to lucky circumstances. Thanks
		  to Chris DeLise for discovering it.
		- Added defines for Linux PPC provided by Rich West.
		- Added patches for Ultrix & NetBSD provided by Simon Burge.
		- Made some minor optimisations, giving somewhere above 1%.

		Still have some of Roger Fujii's minor optimisations to include,
		but now I'm going on a short vacation (my first summer vacation
		in 3 years, hurray!) :-)


1999-07-23	Put the latest version online as 083-unstable. Improvements since
		0.82:
		- Added a "-refresh"-switch to set the update rate of the progress
		  indicator (in order to save CPU-time) and set the default 
		  refresh rate to 2 (half the previous one). This gave a speed
		  increase of slightly more than 1% on my machine.
		- Added a define in SYSTEM.H called PRECISE_TIMER. This is a quite
		  dirty hack by me to allow for more precise calculation of encoding
		  time and is only meant as a tool when optimizing BladeEnc. It
		  calculates and prints out encoding time with higher precision,
		  but slows down the encoding process with about 2 seconds for
		  each file. It's not extremely trustworthy either, but enough
		  for my needs.
		- Started to merge in some optimizations provided by Roger Fujii,
		  most notably a very clever replacement of the bladTabValue()
		  function in loop.c which gave a speed increase of about 7% on
		  my K6-2 350. Unfortunatelly this also makes a notably bigger
		  executable since bladetab.c was changed and fleshed out.
		  I still have some of his optimizations left to merge in, 
		  but they won't do more than about 2% together.
		- Probably one or two other minor changes which I've forgotten...

1999-07-23	Fixed a minor thing having to do with compilation on some
		machines. Put online as bladeenc-082-src-stable-2.tar.gz

1999-07-22	Fixed the embarassing skipping header bug in 0.80/0.81 by
		including a FIFO for the headers (by ISO called sideinfo) in
		formatbitstream2.h. Also got a few patches fixing minor bugs
		and compilation problems (WIN32_ALPHA and SCO OpenServer).
		Regarded this as stable and released it right away. I intend to
		concentrate on including the speedups I've been sent and release
		a new, hopefully much faster, version when that has been done.
		There will probably be an unstable branch of this since it includes
		changes to the FFT calculations that I want to be throughly tested
		for possible quality degradation before I release it to the masses.

1999-06-30      Loads of bugs in 0.80 fixed thanks to you guys out there. 
                Released 0.81 as a bugfix release, so that is now the latest
                stable version. Also added a makefile, not exactly as good as
                a fully working Configure/Make/Make Install system, but I 
                haven't gotten any of those some of you sent me working yet.
                I guess I just have to learn a little bit more about it first,
                until then you'll have to do with this simple, 
                platform-dependent makefile.

1999-06-28      Version 0.80 stable released.