Re: just like midi/seq was treated? (was Re: HEADSUP: ibcs2 and svr4 compat headed for history)

On Jul 02, Taku YAMAMOTO wrote:
> On Fri, 2 Jul 2004 11:07:33 -0400
> Mathew Kanner <mat_at_cnd.mcgill.ca> wrote:
> > On Jun 30, Taku YAMAMOTO wrote:
> > > On Wed, 30 Jun 2004 14:37:47 -0400
> > > Mathew Kanner <mat_at_cnd.mcgill.ca> wrote:
> > > > On Jun 28, Paul Seniura wrote:
> > > 
> > > I object to that statement - IT WAS WORKING without major problems
> > > here at least. (-current as of April 2004)
> > > # I have been using YAMAHA's UX-16, BTW.
> > 
> > 	I'm absolutely shocked to hear this.  This is the first report
> > of midi2 working for anybody.  Are you speaking of the sequencer are
> > are you doing raw midi passthrough?  (Tangentally, I'll admit that I
> > tried to stay away from the old model only to end up recreating the
> > wheel in many aspects, esp. in the seq->midi abstraction.)
> 
> Both of two are working here.
> 
> wine (sequencer), playmidi (sequencer), smfplay (raw midi)
> and my MIDI filtering tool (utilizes both of two) - all of them work.

	Good good, I think that users of 4-x will be overjoyed to hear
this.

> 
> > > And, in fact, IT STILL IS WORKING with recent -current, cvsupped
> > > on June 27. The MIDI stuff was salvaged from Attic with my tiny
> > > modifications to catch up the recent activities like dev_t =>
> > > struct cdev* transitions.
> > > 
> > > I'll tar-up the material and make it accessible if requested.
> > > # And, USB-MIDI port, too :)
> > 
> > 	The new version does bring new features like loadable modules,
> > a kernel thread instead of softclock, an easier interface for
> > attaching new devices and it works regardless of what HZ value is
> > used.
> 
> Did you know that midi2 can be made as a module with additional
> several lines?  I've using midi2 as a module for a year.
> # Sorry for my not publishing these works, BTW.

	Uh sure, but it won't unload, that makes it kind of useless as
a module.

> 
> You have to be aware of that both aproaches can't be HZ-independent, too,
> since 1/HZ is the finest resolution of the scheduling.
> Making sequencer stuff pretending HZ=100 is also a Really Bad Thing (TM)
> because, in theory, softwares using MIDI always want the finest.
> 
> OSS spec clearly says relying on HZ hardcoded as 100 is a Bad Thing
> and encourages the use of SNDCTL_SEQ_CTRLRATE instead, if that matters.

	Um, my head isn't really in the space to remeber, but I can
reliably use software with midi3 that doens't work with midi2
(/dev/sequecer, not /dev/music.)  

	--Mat

-- 
	It's impossible to awaken a man who is pretending to be
	asleep.
			- Navajo saying