[Jens M Andreasen] >On tor, 2004-04-29 at 16:11, Chris Cannam wrote: >> On Thursday 29 Apr 2004 2:19 pm, Jens M Andreasen wrote: >> > I think I will not be really, totally happy before I see something >> > like: >> > >> > void (*midi_msg)(LADSPA_Handle instance, >> > unsigned byte msg[4]); >> >> This would be exactly equivalent to the existing run_synth(), would it >> not? > >Mmm .. This would replace, I think, all of your considerations on how to >extend LADSPA. It is (probably?) the only extension really needed. MIDI >is a very powerful protocol, and the beuty is that the client needs only >to implement those parts it needs to get going.
as i understand it, this is not enough. the run_synth() method allows you to run the synth for a continuous audio block with [MIDI] events interspersed at arbitrary timestamps, while with midi_msg() you'd have to split audio blocks to match event timestamps, which is messy and performance-unfriendly. tim