Jonathan Wilkes escribió:
Maybe there should be a third inlet for voice stealing.
Should one need that, it wouldn't be difficult to create an abstraction to implement that by encapsulating two [poly]s, one with voice stealing and the other without it, switching between them (*).
On the other hand an inlet for dynamically changing the number of voices would probably be more useful (and whatever the use one may find for it, it couldn't be done on top of the static poly)
(*) ohhh, no, well, it _would_ be difficult if you don't assume that there are no pending note_off's when the voice-stealing-activation is changed; it would probably be almost as difficult as re-implementing [poly] from scratch in Pd, if not more...