Analogue-sequencer

> Well, the new directional modes are great - and you needn't worry
> about the processing overheard - at least if my initial tests are
> anything to go by. Just ran all 8 tracks, many of them using random or
> brownian, often with skip incorporated and the timing was still good.
> I had one track with all steps but two skipped and set to random and
> it felt just as punchy as when only the one track was running.

I've been doing some testing myself - I finally added some code to
measure the peak processing time taken by the clock_pulse handler.
There are no timing issues in getting the note messages out - with 8
tracks playing it's taking much less time to generate the outgoing midi
data than it takes to transmit it serially - about 3ms maximum. 8 midi
note-ons require at least 5.5 ms to send, or more like 7.5 ms if they're
not all on the same channel (no running status). 
The problem was with the time it takes to then calculate the next step
for each sequence - this is calculated at the last clock pulse of each
beat. In random mode, it has to work out the effective length of the
pattern (counting non-skipped notes up to the last step), then pick a
random number 'n' in this range, then find the position of the 'n'th
unskipped note.
The current beta, running 8 tracks all with patterns on random is taking
enough time to do this to cause a problem at tempos any higher than 145
bpm - i.e. sync will slip because the next clock_pulse has occured
before the previous one is finished processing.
I did some re-working of the code last night, and have reduced the
overhead significantly. It should be fine well beyond the 240 bpm
maximum.
I'll do some further optimising anyway, as I need to keep things as
quick as possible so I can add other processing.
Anyone needing to run 8 tracks in random mode at tempos greater than 240
bpm probably needs to have a critical listen to their musical output...
;-)

> Also tried experimenting with a track just sending controllers, its
> motion random and using another track for notes, getting those lovely
> "Polymorph" sequences was a dream...

Now that's a neat idea !

Cheers,
Colin f