Re: [Rosegarden-devel] Port/MidiDevice direction ?

[Bob H. <ra...@ba...>]

On Thu, 2003-02-27 at 15:32, Chris Cannam wrote:
> Bob Ham wrote:
> > I tried to find stuff in the archives about this as suggested by Mark,
> > but working with sf.net's mail archives is a nightmare and I couldn't
> > find any mention of it.
> 
> Well, there's quite a lot of related flamewar content accessible
> through the front page of the February list archives -- see the
> messages under the subject "A thought on the Studio stuff - ports vs.
> clients".  Generally "Studio" is quite a good word to search for,
> or would be if SourceForge's search facility wasn't quite so bad.

I had a look through all of these threads, but there didn't seem to
anything new except layouts of dialogs and how to name different alsa
clients.  This is a the major problem in my eyes.

*Why* does rosegarden need a model of the midi network?  If rosegarden
depends on a model of the midi network, which *will* change over
invocations, it will make automating sessions impossible.  There is no
need for rosegarden to pay any attention to its midi connections.  I'll
say that again, because it's very important: rosegarden does *not* need
to pay any attention to its midi connections.  This is the job of
something else.  IMHO, rosegarden's job is to get events from midi (or,
in fact, audio) tracks to their assigned *rosegarden* port.  *Not* to
ports external to rosegarden.  

The ability to define midi instruments is wicked.  That's not the
problem.  The issue is that you're applying these definitions to
*external* alsa clients.  Rosegarden should not know of, nor present to
the user, anything outside of its own midi ports.  Doing so totally
destroys any attempt at automating sessions.

Hope I'm making some sense here,

Bob

-- 
Bob Ham <ra...@ba...>