[PD] Problem with Commercial OSS and PD
Joseph A. Sarlo
jsarlo at mambo.peabody.jhu.edu
Sat Jan 12 17:52:13 CET 2002
I don't have the original so I can't make a diff, but I've attached a file
that has the changed linux_open_midi function from s_linux.c (version
0.34-2, I think). It basically just checks to see that you are only
opening one midi device and that that device is the same. If so, it opens
it O_RDWR. Otherwise it falls back to the normal PD behavior.
PS This is a quick and dirty hack I did for myself and is not meant as a
real solution.
--
______________________________
|
| Joseph A. Sarlo
|
| jsarlo at mambo.peabody.jhu.edu
|______________________________
On Fri, 11 Jan 2002, Timothy Rolls wrote:
> Hi Joseph,
> I was wondering if you could explain your hack of s_linux.c? Or maybe
> e-mail me the file so I could look at it. I'd be really interested in trying
> this before I write to OSS tech support. I'm not too sure on doing this
> myself with out some instruction since I haven't done any C programing in
> about 15 years. Thanks.
>
> Regards,
> Tim
>
> ----- Original Message -----
> From: "Joseph A. Sarlo" <jsarlo at mambo.peabody.jhu.edu>
> To: "Timothy Rolls" <trolls at binghamton.edu>
> Cc: <msp at ucsd.edu>; "Ben Bogart - FMPM/F1999" <bbogart at acs.ryerson.ca>;
> "Pure Data" <pd-list at iem.kug.ac.at>
> Sent: Thursday, January 10, 2002 6:54 PM
> Subject: Re: [PD] Problem with Commercial OSS and PD
>
>
> > I don't know if this issue is related, but I experienced identical
> > symptoms. My problem turned out to be that PD makes two separate calls to
> > open(), once for read and once for write when opening the MIDI device. My
> > driver (serial midi) apparently didn't support this. I just hacked
> > s_linux.c to use one open with O_RDRW and everything has been fine.
> >
> > --
> > ______________________________
> > |
> > | Joseph A. Sarlo
> > |
> > | jsarlo at mambo.peabody.jhu.edu
> > |______________________________
> >
> >
> > On Thu, 10 Jan 2002, Timothy Rolls wrote:
> >
> > > Ok, I've found something interesting. I can either have a MIDI in or
> MIDI
> > > out in PD. If I open with pd -midiindev 0 -midioutdev 1, then I can play
> > > midi out, but not in. If I do the reverse, then I get midi in but no
> midi
> > > out.If I try -midiindev 1 -midioutdev1, I get a message that midiout
> could
> > > not be opened.
> > >
> > > I still question their claim of 100% compatibility with OSS/Free.
> > >
> > > Tim
> > >
> > > ----- Original Message -----
> > > From: "Miller Puckette" <mpuckett at man104-1.ucsd.edu>
> > > To: "Timothy Rolls" <trolls at binghamton.edu>
> > > Cc: "Ben Bogart - FMPM/F1999" <bbogart at acs.ryerson.ca>; "Pure Data"
> > > <pd-list at iem.kug.ac.at>
> > > Sent: Thursday, January 10, 2002 12:36 PM
> > > Subject: Re: [PD] Problem with Commercial OSS and PD
> > >
> > >
> > > > Hmm, I've been getting MIDI out on Free OSS, funny it doesn't work
> > > > on commercial...
> > > >
> > > > cheers
> > > > Miller
> > > >
> > > > On Thu, Jan 10, 2002 at 01:17:35PM -0800, Timothy Rolls wrote:
> > > > > Well, I guess that explains it. But I am curious about the
> legitimacy of
> > > > > their claim that it is 100% compatible with OSS/Free
> > > > > (http://www.opensound.com/linux.html). I am still glad I got it. The
> > > audio
> > > > > quality on my SB PCI 128 is greatly improved. Thanks for the reply.
> > > > >
> > > > > Regards,
> > > > > Tim
> > > > >
> > > > > ----- Original Message -----
> > > > > From: "Ben Bogart - FMPM/F1999" <bbogart at acs.ryerson.ca>
> > > > > To: "Timothy Rolls" <trolls at binghamton.edu>
> > > > > Cc: "Pure Data" <pd-list at iem.kug.ac.at>
> > > > > Sent: Thursday, January 10, 2002 5:59 AM
> > > > > Subject: Re: [PD] Problem with Commercial OSS and PD
> > > > >
> > > > >
> > > > > > I'm using OSS/Commercial and I too only have MIDI In and no MIDI
> out.
> > > > > > This is because OSS does not support MIDI out because most card do
> not
> > > > > > have hardware MIDI out. If MIDI out works with your card with
> other
> > > > > > programs thats because the MIDI translation is done in software,
> not
> > > > > > by the card. This is what I got from OSS/Commercial documentation.
> > > > > >
> > > > > > no sure about a fix, I only use MIDI in.
> > > > > >
> > > > > > Ben
> > > > > >
> > > > > >
> > > > > >
> > > > > > On Thu, 10 Jan 2002, Timothy Rolls wrote:
> > > > > >
> > > > > > > Greetings,
> > > > > > > Yesterday I upgraded from OSS/Free to commercial OSS. Now
> MIDI
> > > > > doesn't
> > > > > > > work in PD, and apparently only PD. All other programs with MIDI
> > > still
> > > > > work
> > > > > > > (Brahms, Jazz, Csound). Is there anyone else using commercial
> OSS
> > > and PD
> > > > > > > with this problem? BTW, I get the message "opened 1 MIDI input
> > > device
> > > > > and 0
> > > > > > > MIDI output devices." This never happened under OSS/Free.
> Thanks.
> > > > > > >
> > > > > > > Regards,
> > > > > > > Tim
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > B. Bogart
> > > > > > ---------
> > > > > >
> > > > > >
> > > > >
> > >
> > >
> >
> >
>
-------------- next part --------------
void linux_open_midi(int nmidiin,
int *midiinvec,
int nmidiout,
int *midioutvec)
{
/* Joe's code, to open with 0_RDRW flag */
if (nmidiin == 1 && nmidiout == 1 && midiinvec[0] == midioutvec[0])
{
int fd = -1;
char namebuf[80];
int devno = midiinvec[0];
if (devno == 1)
{
sys_setalarm(500000);
fd = open("/dev/midi", O_RDWR | O_NDELAY);
if (sys_verbose)
fprintf(stderr,
"device 1: tried /dev/midi; returned %d\n",
fd);
}
if (fd < 0)
{
sys_setalarm(500000);
sprintf(namebuf, "/dev/midi%2.2d", devno-1);
fd = open(namebuf, O_RDWR);
if (sys_verbose)
fprintf(stderr,
"device %d: tried %s; returned %d\n",
devno,
namebuf,
fd);
}
if (fd < 0)
{
sys_setalarm(500000);
sprintf(namebuf, "/dev/midi%d", devno-1);
fd = open(namebuf, O_RDWR);
if (sys_verbose)
fprintf(stderr,
"device %d: tried %s; returned %d\n",
devno,
namebuf,
fd);
}
if (fd >= 0)
{
oss_midiinfd[oss_nmidiin++] = fd;
oss_midioutfd[oss_nmidiout++] = fd;
}
else post("couldn't open MIDI device %d", devno);
}
/* End Joe's code */
else
{
int i;
for (i = 0, oss_nmidiin = 0; i < nmidiin; i++)
{
int fd = -1;
char namebuf[80];
int devno = midiinvec[i];
if (devno == 1)
{
sys_setalarm(1000000);
fd = open("/dev/midi", O_RDONLY);
if (sys_verbose)
fprintf(stderr,
"device 1: tried /dev/midi; returned %d\n",
fd);
}
if (fd < 0)
{
sys_setalarm(1000000);
sprintf(namebuf, "/dev/midi%2.2d", devno-1);
fd = open(namebuf, O_RDONLY);
if (sys_verbose)
fprintf(stderr,
"device %d: tried %s; returned %d\n",
devno,
namebuf,
fd);
}
if (fd < 0)
{
sys_setalarm(1000000);
sprintf(namebuf, "/dev/midi%d", devno-1);
fd = open(namebuf, O_RDONLY);
if (sys_verbose)
fprintf(stderr,
"device %d: tried %s; returned %d\n",
devno,
namebuf,
fd);
}
if (fd >= 0)
oss_midiinfd[oss_nmidiin++] = fd;
else post("couldn't open MIDI input device %d", devno);
}
for (i = 0, oss_nmidiout = 0; i < nmidiout; i++)
{
int fd = -1;
char namebuf[80];
int devno = midioutvec[i];
if (devno == 1)
{
sys_setalarm(1000000);
fd = open("/dev/midi", O_WRONLY);
if (sys_verbose)
fprintf(stderr,
"device 1: tried /dev/midi; returned %d\n",
fd);
}
if (fd < 0)
{
sys_setalarm(1000000);
sprintf(namebuf, "/dev/midi%2.2d", devno-1);
fd = open(namebuf, O_WRONLY);
if (sys_verbose)
fprintf(stderr,
"device %d: tried %s; returned %d\n",
devno,
namebuf,
fd);
}
if (fd < 0)
{
sys_setalarm(1000000);
sprintf(namebuf, "/dev/midi%d", devno-1);
fd = open(namebuf, O_WRONLY);
if (sys_verbose)
fprintf(stderr,
"device %d: tried %s; returned %d\n",
devno,
namebuf,
fd);
}
if (fd >= 0)
oss_midioutfd[oss_nmidiout++] = fd;
else post("couldn't open MIDI output device %d", devno);
}
}
if (oss_nmidiin < nmidiin || oss_nmidiout < nmidiout || sys_verbose)
post("opened %d MIDI input device(s) and %d MIDI output device(s).",
oss_nmidiin,
oss_nmidiout);
}
More information about the Pd-list
mailing list