[Cs-dev] PortMIDI cleanup
| Date | 2005-11-02 19:20 |
| From | Victor Lazzarini |
| Subject | [Cs-dev] PortMIDI cleanup |
I spotted something that might perhaps be wrong on pmidi.c. It is actually to do with porttime; porttime is started with Pt_Start(), but never stopped with Pt_Stop(). Adding a call to it to the module destructor allows the portmidi module to re-start without an error in porttime. I can't test this, because I don't have a working MIDI, but I can see that the initialisation seems to be happening. Can anyone (Istvan?) add it to the CVS sources, so that David can test it again? I also have no CVS access at the moment. The Portmidi code seems to have global statics everywhere, no wonder it cannot run in more than one instance. Perhaps a request to make it re-entrant could be passed to its developers. Victor ------------------------------------------------------- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2005-11-02 20:22 |
| From | Istvan Varga |
| Subject | Re: [Cs-dev] PortMIDI cleanup |
Victor Lazzarini wrote: > I spotted something that might perhaps be > wrong on pmidi.c. It is actually to do with > porttime; porttime is started with Pt_Start(), > but never stopped with Pt_Stop(). Adding a call > to it to the module destructor allows the portmidi module > to re-start without an error in porttime. Yes, this may indeed be the problem. However, calling Pm_Terminate() amd Pt_Stop() from the module destructor may not actually be a very good idea, as it is always called regardless of whether MIDI devices were opened or not. Perhaps I should register a reset callback after calling Pm_Initialize() and Pt_Start() ? Even then, there are still problems with multiple instances due to the requirement for the global initialize/terminate calls that apparently cannot be nested in PortMidi (not to mention any static variables in the PortMidi sources). ------------------------------------------------------- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2005-11-02 21:39 |
| From | Istvan Varga |
| Subject | Re: [Cs-dev] PortMIDI cleanup |
Victor Lazzarini wrote:
> I spotted something that might perhaps be
> wrong on pmidi.c. It is actually to do with
> porttime; porttime is started with Pt_Start(),
> but never stopped with Pt_Stop().
Well, for now, I committed this changed code:
static int stop_portmidi(CSOUND *csound, void *userData)
{
(void) csound;
(void) userData;
Pm_Terminate();
Pt_Stop();
return 0;
}
static int start_portmidi(CSOUND *csound)
{
if (Pm_Initialize() != pmNoError) {
csound->ErrorMsg(csound, Str(" *** error initialising PortMIDI"));
return -1;
}
if (Pt_Start(1, NULL, NULL) != ptNoError) {
csound->ErrorMsg(csound, Str(" *** error initialising PortTime"));
return -1;
}
return csound->RegisterResetCallback(csound, NULL, stop_portmidi);
}
-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. Download
it for free - -and be entered to win a 42" plasma tv or your very own
Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net |
| Date | 2005-11-02 22:13 |
| From | Michael Worobec |
| Subject | [Cs-dev] a CSound moment at the coffee shop ... |
Hey, I had an interesting moment at the local coffee shop yesterday - I was in line to get a coffee, with my CSound book in hand, planning to read a chapter or two over a latte. The woman behind me spoke up: "The CSound book!" It turns out she just finished a grad program at MIT during which she took the CSound class. Cool coincidence. Never leave home without your CSound book... Mike Worobec ------------------------------------------------------- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2005-11-03 03:11 |
| From | Iain Duncan |
| Subject | Re: [Cs-dev] PortMIDI cleanup |
Which platform are you on Victor? You should be able to test portmidi with virtual midi loopbacks of some sort. I have used PD and virmidi on linux before to test portmidi code of my own, and on windows I know there is midi-yoke/midi-ox. Iain Victor Lazzarini wrote: > I spotted something that might perhaps be > wrong on pmidi.c. It is actually to do with > porttime; porttime is started with Pt_Start(), > but never stopped with Pt_Stop(). Adding a call > to it to the module destructor allows the portmidi module > to re-start without an error in porttime. I can't > test this, because I don't have a working MIDI, but > I can see that the initialisation seems to be > happening. > > Can anyone (Istvan?) add it to the CVS sources, so > that David can test it again? I also have no CVS > access at the moment. > > The Portmidi code seems to have global statics everywhere, > no wonder it cannot run in more than one instance. > > Perhaps a request to make it re-entrant could be passed > to its developers. > > Victor > > > ------------------------------------------------------- > SF.Net email is sponsored by: > Tame your development challenges with Apache's Geronimo App Server. Download > it for free - -and be entered to win a 42" plasma tv or your very own > Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > ------------------------------------------------------- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2005-11-03 13:46 |
| From | Istvan Varga |
| Subject | Re: [Cs-dev] PortMIDI cleanup |
Note that there are also native drivers for ALSA and Windows MME (but not OS X yet), and these are expected to work correctly when restarted or run in multiple instances. I did not test these extensively (particularly the Win32 one), so feedback on how well they work is welcome. The use of ALSA is enabled with: -+rtmidi=alsa -M hw:x,y where x and y correspond to /dev/snd/midiCxDy for MME, use these options: -+rtmidi=mme -M x or -+rtmidi=winmm -M x where x is a device number (not necessarily the same as with PortMidi). An out of range number like -M 1000 will print the list of available devices. Of course, MIDI output is also supported, just use the -Q option (device selection syntax is the same as in the case of input). Iain Duncan wrote: > Which platform are you on Victor? You should be able to test portmidi > with virtual midi loopbacks of some sort. I have used PD and virmidi on > linux before to test portmidi code of my own, and on windows I know > there is midi-yoke/midi-ox. > > Iain > > Victor Lazzarini wrote: > >> I spotted something that might perhaps be >> wrong on pmidi.c. It is actually to do with >> porttime; porttime is started with Pt_Start(), >> but never stopped with Pt_Stop(). Adding a call >> to it to the module destructor allows the portmidi module >> to re-start without an error in porttime. I can't >> test this, because I don't have a working MIDI, but >> I can see that the initialisation seems to be >> happening. >> >> Can anyone (Istvan?) add it to the CVS sources, so >> that David can test it again? I also have no CVS >> access at the moment. >> >> The Portmidi code seems to have global statics everywhere, >> no wonder it cannot run in more than one instance. >> >> Perhaps a request to make it re-entrant could be passed >> to its developers. ------------------------------------------------------- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2005-11-03 16:39 |
| From | David Akbari |
| Subject | Re: [Cs-dev] PortMIDI cleanup |
On Nov 3, 2005, at 8:46 AM, Istvan Varga wrote: > Note that there are also native drivers for ALSA and Windows MME > (but not OS X yet), and these are expected to work correctly when > restarted or run in multiple instances. CoreMIDI also supports restarting and running in multiple instances; I would imagine this is the OS X equivalent for the native ALSA and Windows MME drivers. In this situation it is a shame ma++'s MacCsound is not open source, as I am fairly certain that it implements a CoreMIDI virtual device driver and of course all of it's OS specific inter-application bussing features. The main disadvantage however is that you cannot explicitly specify *which* device on your system to use (ie. MacCsound does not implement an equivalent to -M), in favor of a preference pane from another application which is also outputting data to the inter-application bus. Thus, users are forced to get the NSApplicationServices.framework involved. In any event, I have tested the new cstclsh shell in OS X from a fresh Nov 2 CVS build and it is possible to re-instantiate PortMIDI in multiple instances without problems. I was also able to run the attached example several times successively without incident. I used the following commandline: % csCompile -odac:1 -idevaudio:1 -d -+noninterleaved=1 -+rtaudio=CoreAudio -M0 /Users/daveakbari/Desktop/midiTest.csd with the following unified CSD example (just for testing): |