amp0.7 TODO file
================

###---> 0.8

check if EOF when gethdr() doesn't return 0, if not, indicate synchronization error

redo antialiasing butterflies to use less multiplications (use ptr. arithm)

no_of_imdcts is not calculated with requantize_mono, just set to 32. fix it. 

when short blocks are used, non_zero for the channel is frequently just slightly bigger
than sfb boundary, and that part is usually all zero. detect this so that we don't do
any unneded imdcts.

improve no_of_imdcts calculation for short blocks

let the counter display minutes:seconds instead of a frame count

hard code the huffman tables into huffman.h - the way it is now is very inconvenient.

an altogether better huffman decoding, this one is untouched since the first drafts in apr96

hold some tmp values in _poly/dct/adding, enhance the damn thing

implement parse-only mode - should be useful, also quality debug output

cut down t_43 to 256 values, and for higher use iis-proposed algorithm

move printout to audio.c or somewhere

clean up wave header writing

clean up the makefile

clean up the player - byte swapping policy, (mixer support?), make it use non-blocking writes...

layers 1&2

collect all the global variables into a layer3_globals struct and pass pointers, so we can decode 
more streams in parallel  

instead of mucking about with exporting res[][][] and s[][][] so that we can acces them from layer3.c
make poly() pick the needed elements out of res[][][] itself, and a special zero-out function for
res and s

###---> 0.9

a GUI

CRC checking

Netscape helper application (look into that)

impose a limit on processor power spent

build binary and RPM distributions

###---> 1.0

nothing special, just make it a good, reliable piece of software
