Open hagenw opened 7 years ago
The same problem applies to ssr_brs_wfs()
.
I would not want to change how get_ir()
works either. But I have the following suggestion: The only delay that should be returned from ir_correct_distance()
is the one that can be caused by delayline()
, I think. So this should be returned from ir_correct_distance()
to get_ir()
to ir_generic()
where it can be added to the delay returned by driving_function_imp_wfs()
. What do you think?
Ah ok, now I see the real problem. I thought it was just a problem performance wise. But the problem is of course that delayline()
can add a processing delay already inside ir_correct_distance()
. So the only solution seems indeed to return that delay in ir_correct_distance()
and pass it on.
I do not like that very much as it add a returning parameter to get_ir()
that you would not expect there, but I have no better proposal at the moment.
This is a follow up on #115.
There @VeraE wrote that
delayline()
gets called several times:One way to improve this might be to let
ir_correct_distance()
return delay and weight instead of the actual shifted impulse response. The downside is that this would complicate howget_ir()
works as this would then need to return the unmodified impulse response and the delay and weight. I guess in this case it would even be better to removeir_correct_distance()
fromget_ir()
and apply it somewhere else.Maybe we find a good solution for this, but in my opinion it would be also ok, to stay with the current state.