« Return to Thread: MIDIClient.init broken on linux

Re: MIDIClient.init broken on linux (rethinking for osx and windows?)

by nescivi :: Rate this Message:

Reply to Author | View in Thread

Hiho,

I committed a fix for now for Linux,
but I really would like to get this issue properly looked into...

So the question:
Is it possible with CocoaMIDI and PortMIDI to just initialize the client,
without specifying the number of ports?

If so, then we could create the ports, after initializing the MIDI service,
and everything would be much cleaner codewise...

sincerely,
Marije

On Sunday 25 May 2008 19:15:15 nescivi wrote:

> Hiho,
>
> On Monday 05 May 2008 12:28:10 you wrote:
> > I don't think its needed at all.  you would never wish to limit what you
> > connected to.  the only reason it was like that was because the whole
> > midi thing was copied from the API for cocoa.
> > but I would have to look at the source code, the cocoa stuff.  its a bit
> > snakey.
>
> any further thoughts about this?
> If this could be changed to have only the number of ins and outs be
> specified when using connect, so, after the list is retrieved.
>
> sincerely,
> Marije
>
> PS, isn't this an 80's technology? we might as well have been writing
> physical letters to and fro about this issue ;)
>
> > On Mon, May 5, 2008 at 6:01 PM, nescivi <nescivi@...> wrote:
> > > How essential is it for MacOSX (and Windows) to define the number of
> > > in and outport upon init, if they are later retrieved for connecting
> > > with list?
> > > I am wondering whether it is strictly necessary to specify the number
> > > of ports when initialising SC as a MIDI client.
> > >
> > > sincerely,
> > > Marije
> > >
> > > On Mon, May 5, 2008 at 11:19 AM, felix <felix@...>
wrote:

> > > > On Sun, May 4, 2008 at 11:52 PM, nescivi <nescivi@...> wrote:
> > > > > On Sunday 04 May 2008 09:44:58 felix wrote:
> > > > > > what changed in that checkin is that it calls MIDIClient.list
> > > > > > which
> > >
> > > is
> > >
> > > > the
> > > >
> > > > > > correct way to find out how many ins/outs you have.
> > > >
> > > > (on os x)
> > > >
> > > > > So the MIDI philosophy on Linux is slightly different then on Mac
> > > >
> > > > right.
> > > >
> > > > I wonder what the temporary solution should be to get  MIDIIn.init
> > >
> > > working
> > >
> > > > again for linux.
> > > >
> > > > should we put a platform check directly into MIDIIn.init ?  and then
> > >
> > > what
> > >
> > > > can be done ?  perhaps just skip the .list and go straight to connect
> > > > initialize.  this just means that on linux you have to specify how
> > > > many
> > >
> > > ins
> > >
> > > > and outs you want.
> > > >
> > > > sorry, but I'm not in much of a position to suggest the best
> > > > solution.
> > > >
> > > > for those wondering:
> > > > the bug I was fixing was to make MIDIIn.init automatically connect to
> > >
> > > all
> > >
> > > > possible os x ports.
> > > >
> > > > > A possible solution would be to add a primitive to initialize just
> > > > > the connection with the sequencer without opening any ports. After
> > > > > that
> > >
> > > any
> > >
> > > > other
> > > >
> > > > > methods can be used.
> > > > >
> > > > > sincerely,
> > > > > Marije
>
> _______________________________________________
> Sc-devel mailing list
> Sc-devel@...
> http://lists.create.ucsb.edu/mailman/listinfo/sc-devel


_______________________________________________
Sc-devel mailing list
Sc-devel@...
http://lists.create.ucsb.edu/mailman/listinfo/sc-devel

 « Return to Thread: MIDIClient.init broken on linux