Re: midi port names

On Wed, 2008-10-01 at 21:05 +0200, Paul Davis wrote:
> On Wed, 2008-10-01 at 20:56 +0200, Ralf Mardorf wrote:
> > And why isn't it possible to autodetect the clients? The MIDI devices
> > and soft synth have names. ALSA can auto detect them, resp. gets that
> > information by the applications or drivers and why is this less portable?
> 
> we can do all that. its less portable because the device names may
> change but still be functionally the same. do you really want your
> sessions to no longer load when you replace (one of) your MIDI
> interface(s) ?
> 
> what i actually think would be better is to define a new convention for
> patching clients. JACK already supports port aliases, so a port can have
> up to 3 different names. We could easily define two names (the canonical
> name and 1 alias) based on a fixed numbering and ALSA sequencer names.
> What would be needed is a way to switch between which names are
> displayed. Then everyone is happy. I think.

Using system level names as user pretty names always hits this problem eventually...

Jack needs a simple key/value metadata system, and 'pretty names' for
ports should be handled by that.  Along with auto-connect and about 19
thousand other long standing jack issues...

Anyway, "fixed numbering" is really the problem here, as far as midi
goes (or, in general, multiple interfaces).

Port 2 means /nothing useful/ on almost any system more advanced than a
single sound card with one MIDI interface.  It must be Port 2 _on device
x_

Aliases aside, this is the problem that needs fixing for e.g. stored
connection information to actually be portable.  On my system my 'real'
MIDI interface ports start at 2.  Unportable to the max.

-dr