22 Nov
2018
22 Nov
'18
10:12 p.m.
On 11/22/2018 07:28 PM, David Runge wrote:
Rabbit hole stuff! SuperCollider came to a similar
conclusion:
https://github.com/supercollider/supercollider/issues/4116
This is a different issue. In SuperCollider's case they do want NaN and
not finite-math, that is not usually the case for audio.
If you have perform a calculation that can produce NaN or infinite
values or needs to distinguish +/-0 in your DSP code, there is a much
bigger issue.
ciao,
robin