[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Sc-devel] status: audio rate demand ugens

I've found a provisional fix for forcing audio rate inputs into the ugen graph if they are not used elsewhere. I've committed this (rev 7106) until it is fixed elsewhere.

There is still one weird set of bugs only present in binary operators with demand rate ugens. These bugs have nothing to do with the above fix (they were there before).

This is a bit hard to track. When you add a demand ugen with an audio rate input and one with none audio rate input, the result is about 1/3 of what it should be! The binary op method in the plugin is called, but it is passed a wrong _input_. This input comes from the audio rate unit, but seems to be scaled for some reason. This could actually come from an interpolation of the input that escapes me.

I've attached the tests.

audio + Dseq([0, 0], inf); // ok

Dseq([audio, audio], inf) + Dseq([0, 0], inf); // not ok

Dseq([audio, 0], inf) + Dseq([0, audio], inf); // but this is ok!

some values from the buggy examples (input / demand)

0.745267 / 0.307983

0.50351 / 0.192182

-0.164077 / 0.0103039

ok, I've committed this.

Thanks. Works fine for me. I can't see any reason not to commit it,
apart from the known and unsolved issue, which we know the workaround
for; so I say commit.


2008/1/20, Julian Rohrhuber <rohrhuber@xxxxxxxxxxxxxx>:
  >Hm, can't get the patch to apply. Could you send it as an
attachment please?



Attachment: Demand_audio_rate.rtf
Description: Binary data