On 07/12/2010 20:10, Kassen wrote:
One thing I'd be interested in is using this to sync Fluxa's "(seq )"
to other musicians. For that kind of thing we might also need
something like a timestamp of when the actual message was received and
probably a "latency offset" that we could set on a per-system basis in
.fluxus.scm
Interesting. All MIDI messages are time-stamped by RtMidi. The time
stamp is not used by fluxus' MIDI listener, but it's possible. We have a
few problems here:
- what's the meaning of this timestamp?
- is this meaning platform-dependent?
- how can we convert this time stamp to a value that can be compared to
(time)?
I need to experiment...
I've also found what I think is a bug in RtMidi :
Not sure how long ago you send this. In my experience Gary does reply
to such issues but sometimes it takes a bit.
11 or 12 days...
I do find this a bit of a odd error to make by mistake. Couldn't it be
that rtMidi was assuming all messages are three bytes? I'm asking
because I know that ChucK also uses rtMidi and does that. Sending 1
byte messages is possible there (not receiving) but you have to use
two padding bytes. That might, of course, be a ChucK issue, not a
rtMidi one. I commented on that several times and proposed some
possible solutions but didn't get a clear reply.
This may be a copy-paste error from another part of the code:
else if ( status < 0xF3 ) {
if ( status == 0xF1 && (data->ignoreFlags & 0x02) ) {
// A MIDI time code message and we're ignoring it.
size = 0;
iByte += 3;
}
else size = 3;
}
And not all messages are 3-byte-long:
if ( status < 0xC0 ) size = 3;
else if ( status < 0xE0 ) size = 2;
else if ( status < 0xF0 ) size = 3;
else if ( status == 0xF0 ) {
Mysterious.
At any rate; good job!
Thanks!
--
G.
