karl@sugar.UUCP (Karl Lehenbauer) (05/01/88)
(I offer this information as a summary of some MIDI hassles I've
been having. It might be of use to others trying to do MIDI stuff on
the Amiga.)
I have an application I'm working on where I'm downloading data from
a sampling synthesizer. It uses a handshaked protocol of system
exclusive messages. I'm using Bill Barton's midi.library to handle
the actual port I/O.
Since I started on it yesterday evening, I'd been compiling on the 2000
and carrying a disk over to the 1000 to try it since I didn't have a
MIDI interface for the 2000. Well, that got tiring, so today I bought
an ECE MIDI interface for the 2000. Here's the rub. My software hardly
ever have to do retries on the 1000; it would transfer 2000 blocks
of 136 bytes each without a single error. On the 2000, though, the code
had to do a retry aboute one per 25 blocks transferred, or about 80 for the
case of transferring 2000 blocks.
I set around and tried to zen it out. I figured it was either hardware
or software. :-) On the assumption that there were extra tasks chewing
time on the 2000, I pulled out the extremely useful little tool, "stat", that
Bill included in the MIDI library that can print various things about the
system, such as lists of CLIs, tasks, MIDI ports, MIDI tasks, open
libraries, and such. There were a lot of tasks. I tried booting the 2000
with the same workbench disk I used on the 1000, thus no hard disk mount,
no hddisk.device and DH0 tasks. DH0 incidentally runs at the same priority
as the MidiIn and MidiOut tasks, 10. Well, that's what it is. I've only
tried one transfer, but I only had two retries out of 2000 packets.
In his midi.library docs Bill asks if anyone has a streamlined midi.device
that works better than the serial device because the serial device loses
bytes even with SERF_RAD_BOOGIE set. Does anyone? How about in 1.3?
--
"I think Michael is like litmus paper - He's always trying to learn" - Liz
Taylor ..!{bellcore!tness1,uunet!nuchat}!sugar!karl, Unix BBS (713) 438-5018doug-merritt@cup.portal.com (05/03/88)
Karl mentions MIDI problems on the 2000 vs the 1000, and asks if
the serial device will be improved under 1.3.
As it happens, Bryce Nesbitt, now working for CBM in PA, is working
on this. He says that there will be a few improvements for 1.3,
and a lot under 1.4. I don't know any details at all.
However, it stands to reason that if you MIDI guys out there make
your wants and needs known, there will at least be a chance that Bryce
will be able to eventually accomodate them. Whereas if you don't...
Doug Merritt ucbvax!sun.com!cup.portal.com!doug-merritt
or ucbvax!eris!doug (doug@eris.berkeley.edu)
or ucbvax!unisoft!certes!doug