[iiwusynth-devel] RE: Command response framing and TCP/IP

[M. Nentwig]

> Lets put off the dedicated socket until someone asks.  Implementing
support for
> multiple sockets will force the multithreaded command processor
feature.

The 'realtime data socket' was just an idea to ease processing on the
application side. Forget about it, if it's inconvenient.
If we implement it, the idea is that 'realtime' commands go to a
separate socket, to avoid mixing up noteon/noteoff/cc/... events with
output from a command like 'inst 0'.
Also, if the connecting application does not want to know about realtime
issues, it just doesn't connect to the realtime socket and that's it.
I think output should always be done on normal priority, otherwise
dropouts can result (try --verbose to stdout, I observe regular
dropouts, probably because the MIDI thread that triggers the output runs
on raised priority).

But regarding threads I end up with the same problem anyway - IMO the
command processor must be able to cope with queries from different
threads:
-sdin (shell)
-from MIDI sysex
-TCP/IP

 The solution would probably be to have an own thread for the command
processor, that processes queries in a FIFO manner.