RE: [AD] Modified color.c and mixer.c

[ Thread Index | Date Index | More lists.liballeg.org/allegro-developers Archives ]


> What happens if we do overflow?  I'd imagine it wraps at the
> moment, which is nasty.  Whatever we do, there would be an
> artifact, but maybe clamping would sound better (or less bad).
> We can't implement that if it's slow though.

That's something I considered mentioning before, but I thought the message
would be too long. The mixer actually uses a clipping table already - and
what's more, the clipping table is only enabled if more than 8 voices are
reserved. It's as if it was initially designed the way I've just programmed
it.

I should like to point out that when I compile in Windows, using DirectX,
the sound output is much greater - and it does distort. Although this
probably varies between computers, I think it would be good to keep the
platforms as similar as possible. Then the programmer can make sure it
doesn't distort, without having to play an important sample on 8 channels at
once :-)

Can someone tell me what happens in other OSs, such as Linux?

I shan't send anyone the code for mixer.c yet, until we think this matter is
resolved. I have, however, sent color.c to Bob. He has kindly offered to do
the patches for me. Besides, there's another small problem with looped
quality=2 resampling, which I might try to fix at some stage if it doesn't
make it too slow.

Shawn, what do you think about mixer.c? I'd like to get your opinion
first-hand.

Ben Davis



Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/