-
-
Notifications
You must be signed in to change notification settings - Fork 0
/
TODO.txt
41 lines (39 loc) · 1.76 KB
/
TODO.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
- Resamplers need altivec versions.
- Resamplers need to suck less.
- Vorbis extension needs resampling to device format.
- Vorbis extension missing altivec mixer for mono output.
- Vorbis extension should cope with alignment to mix with Altivec whenever
humanly possible.
- Capture extension
- Pitch resampling.
- Mono output.
- > 2 channel output.
- Make alExtension.c not suck.
- Device enumeration feels hacky.
- Look for FIXMEs.
- Remove ov_read_callbacks abstraction and use ogg buffer directly to
relieve memcpy pressure and reduce memory bandwidth?
- Make sure Ogg Vorbis always aligns buffers on 16 byte boundaries.
- Make altivec stereo16 mixer works with unaligned input (and use scalar for
unaligned output at start of buffers, so we can avoid ever falling into a
completely scalar mixing path. See mono mixers for reference.
- Altivec mixer for stereo8 is missing.
- Arbitrary downsampling is broken/slow.
- Port to Linux.
- How to render stereo sounds on > 2 channel setups?
- Need test suite for mixers and resamplers.
- Need documentation for various extensions I'm creating on a whim.
- ALUT calls exit() if it can't load a WAV file!
- Source's needsRecalculation must be set if the listener moves.
- Spec compliance testing framework...?
- Vorbis doesn't resample to device format before mixing!
- Consolidate all the isInput/isOutput crap in alContext.c ...
- Make all these new extensions use sane values their enums.
- Can we get more efficiency out of capture by seeing if the
device is already being used for output and latching on to the mixer
callback?
- The ring buffer is crap.
- There's a lot of room for optimization in the capture callback.
- There's a lot of untested codepaths in the capture extension.
- 80-column source code cleanup.
// end of TODO ...