After a lot of travel, I'm back at work on PortMidi, especially for the
sysex problems that have been reported. After trying many things, I
produced the following code where I take what seems to be a perfectly good
MIDIHDR pointing to sysex data. Note that the "DEBUG CODE" just copies the
sysex data that's already in place. This works, but if I do everything but
assigning the lpData pointer, it does not work. I'm testing sending to a
MIDI Yoke device (MIDI Yoke is a device driver that creates virtual MIDI
devices allowing applications to interconnect) and monitoring output with
MIDI Ox (which has a MIDI monitor program) as well as with a new
PortMidi-based MIDI monitor.
Here's the code segment:
m->hdr->dwBufferLength = m->sysex_byte_count;
/* DEBUG CODE: */
{ int i; int len = m->hdr->dwBufferLength;
char *newmsg = (char *) malloc(len);
for (i = 0; i < len; i++) newmsg[i] = m->hdr->lpData[i];
m->hdr->lpData = newmsg;
}
m->error = midiOutLongMsg(m->handle.out, m->hdr, sizeof(MIDIHDR));
if (m->error) rslt = pmHostError;
The block labeled DEBUG CODE just makes a copy of the sysex data, which is
already on the heap. I've looked at lpData and the data itself with a
debugger before and after the copy, and I've used the debugger to stop
execution right after the return from midiOutLongMsg() to make sure
nothing is corrupting the data, and everything looks ok. If I comment out
just the "m->hdr->lpData = newmsg" line, this code does not work, so I'm
really baffled. I suspect something else may be going on with the library,
drivers, compiler settings, etc., but I can't imagine how just copying
from one heap location to another could make any difference.
Also, the code seems to work without the heap copy when the device is real
hardware (a MidiMan 1x1 interface), so there seems to be some interaction
with the MIDI-OX driver. I'm sending copies of this to midiox.com and my
neighbor/windows guru, but if you have any suggestions or explanations or
work-arounds (other than allocating and copying everything on the heap),
please let me know.
-Roger
_______________________________________________
media_api mailing list
[email protected]
http://www.create.ucsb.edu/mailman/listinfo/media_api
