Re: [seq24-users] Midi Clock Sync patch (slave seq24 to external sequencers)

[Guido S. <gui...@ba...>]

Am Tue, 14. Apr 2009 um 21:43:24 -0500 schrieb Kevin Meinert:

> Hello Everyone,

Hi Kevin,

> Attached is a patch to upgrade perform.cpp and perform.h to have the Midi
> Clock Sync feature.   Test it out and tell me how it goes.

[...]

> documentation:
> ==============================================================
> Midi Clock input:
> 
> seq24 will slave off of incoming MidiClock messages
> 0xfc 0xfa 0xf8   (start stop and tick)
> 
> This has been tested with another seq24 instance (stop/start works)
> This has been tested with a x0xb0x in (start/stop works)
> 
> It will also pay attention to 0xfb (continue) and 0xf2 (Midi Song Pos) but
> these aren't tested...

I would like to come back to this topic. Some weeks ago I followed a
discussion on the Linux Audio User (LAU) mailinglist, dealing with MIDI
clock support:

http://lists.linuxaudio.org/pipermail/linux-audio-user/2009-May/060534.html

Reading our code I found the current implementation in seq24 is not at a
perfect state. I spend some time to look for an internet resource about
this topic and found this article, giving some clear statements about
how a "MIDI clock client" should behave:

  http://home.roadrunner.com/~jgglatt/tech/midispec/seq.htm

The current code is not so far away from the theory, but there are some
basic differences. As I understand the MIDI events should trigger
actions like this:

MIDI Start:
  * Prepare for a start at the very beginning of the song
  * if running, ignore this event

MIDI Continue:
  * Prepare for a start at the current position
  * if running, ignore this event

MIDI Clock:
  * if not running, start running

MIDI song position:
  * if running, ignore event else update position

MIDI Stop:
  * if running, stop else ignore event

Do you agree?
Guido


-- 
http://www.bayernline.de/~gscholz/
http://www.lug-burghausen.org/