Thank you, this is helpful.  But to me it seems then no to be so much a
problem with MIDI, as with the fact that in real-time applications you
don't know in advance what the duration will be.  Any streaming protocol
would be in trouble or am I missing something?

Job van Zuijlen  

Paul Barton-Davis wrote:
> 
> 
> so, yes, a Csound score has some semantics, because its very
> definition includes a specification of voice duration that is absent
> is a MIDI data stream.
> 

>Thank you, this is helpful.  But to me it seems then no to be so much a
>problem with MIDI, as with the fact that in real-time applications you
>don't know in advance what the duration will be.  Any streaming protocol
>would be in trouble  ....

i think so, yes.

you know, in some ways, this goes to the heart of the difference
between improvisation-as-composition and composition itself. yes, a
scorefile lets you do things that no streaming protocol could. but
realtime control of a physical device connected to (or even IS-A)
sound source (not necessarily, and perhaps even preferably not, via
MIDI) lets you do a different set of things than the scorefile. the
differences in the two approaches should be seen as complimentary
features, not conflicts. its just too bad that Csound is modelled
solely on the compositional process, not the improvisational one. on
the other hand, this creates a nice niche for quasimodo, grainwave,
supercollider etc.