22 Apr
2011
22 Apr
'11
10:45 p.m.
Hello all,
after reading ASLA docs for some hours I've not found an
answer to the following:
Given the ALSA sequencer client and port _names_, find the
numbers required to set up a connection (a 'subscription'
in ALSA lingo).
Anyone knows how to do this ?
Ciao,
--
FA
22 Apr
22 Apr
10:51 p.m.
On Fri, Apr 22, 2011 at 4:45 PM, Fons Adriaensen <fons(a)linuxaudio.org> wrote:
Hello all,
after reading ASLA docs for some hours I've not found an
answer to the following:
Given the ALSA sequencer client and port _names_, find the
numbers required to set up a connection (a 'subscription'
in ALSA lingo).
ALSA MIDI ports have names?
11:10 p.m.
On Fri, Apr 22, 2011 at 04:51:53PM -0400, Paul Davis wrote:
On Fri, Apr 22, 2011 at 4:45 PM, Fons Adriaensen
<fons(a)linuxaudio.org> wrote:
They do. For example:
fons@zita1:~> aconnect -i
client 0: 'System' [type=kernel]
0 'Timer '
1 'Announce '
client 14: 'Midi Through' [type=kernel]
0 'Midi Through Port-0'
client 20: 'TerraTec EWS88MT' [type=kernel]
0 'TerraTec EWS88MT MIDI'
client 24: 'MPU-401 UART' [type=kernel]
0 'MPU-401 UART MIDI'
So given e.g. 'TerraTec EWS88MT','TerraTec EWS88MT MIDI'
how do I find 20,0 (from within an application) ?
Or in other words, what is ALSA's get_port_by_name() ?
Ciao,
--
FA
Hello all,
after reading ASLA docs for some hours I've not found an
answer to the following:
Given the ALSA sequencer client and port _names_, find the
numbers required to set up a connection (a 'subscription'
in ALSA lingo).
ALSA MIDI ports have names? 
23 Apr
23 Apr
12:07 a.m.
On Fri, 2011-04-22 at 21:10 +0000, Fons Adriaensen wrote:
On Fri, Apr 22, 2011 at 04:51:53PM -0400, Paul Davis
wrote:
Alsa sequencer names aren't unique or intended for use as identifiers,
you probably don't want to do this if it's at all possible to avoid...
-dr
On Fri, Apr 22, 2011 at 4:45 PM, Fons Adriaensen
<fons(a)linuxaudio.org> wrote:
They do. For example:
fons@zita1:~> aconnect -i
client 0: 'System' [type=kernel]
0 'Timer '
1 'Announce '
client 14: 'Midi Through' [type=kernel]
0 'Midi Through Port-0'
client 20: 'TerraTec EWS88MT' [type=kernel]
0 'TerraTec EWS88MT MIDI'
client 24: 'MPU-401 UART' [type=kernel]
0 'MPU-401 UART MIDI'
So given e.g. 'TerraTec EWS88MT','TerraTec EWS88MT MIDI'
how do I find 20,0 (from within an application) ?
Or in other words, what is ALSA's get_port_by_name() ? Hello all,
after reading ASLA docs for some hours I've not found an
answer to the following:
Given the ALSA sequencer client and port _names_, find the
numbers required to set up a connection (a 'subscription'
in ALSA lingo).
ALSA MIDI ports have names?
12:30 a.m.
On Fri, Apr 22, 2011 at 06:06:54PM -0400, David Robillard wrote:
Alsa sequencer names aren't unique or intended for
use as identifiers,
you probably don't want to do this if it's at all possible to avoid...
If that's true, what purpose do they have at all ?
In this concrete case, I have Python class which acts as
an interface to the Novation Launchpad. I want its ports
to connect to the launchpad, as any other connection
wouldn't make any sense at all. I know the names of
the required ports, but not the numbers.
Ciao,
--
FA
12:55 a.m.
On Fri, Apr 22, 2011 at 6:30 PM, Fons Adriaensen <fons(a)linuxaudio.org> wrote:
On Fri, Apr 22, 2011 at 06:06:54PM -0400, David
Robillard wrote:
for myself, i think its reasonable accurate to say that ALSA's
identification of MIDI ports is totally FUBAR'ed. this doesn't help
much.
i'm not quite sure what you mean about ardour (2's) behaviour, but its
not on my mind much at this point since ardour3 doesn't use ALSA MIDI
for anything at all, and will not do so again.
Alsa sequencer names aren't unique or
intended for use as identifiers,
you probably don't want to do this if it's at all possible to avoid...
If that's true, what purpose do they have at all ?
In this concrete case, I have Python class which acts as
an interface to the Novation Launchpad. I want its ports
to connect to the launchpad, as any other connection
wouldn't make any sense at all. I know the names of
the required ports, but not the numbers.
1:22 a.m.
On Fri, Apr 22, 2011 at 06:55:18PM -0400, Paul Davis wrote:
for myself, i think its reasonable accurate to say
that ALSA's
identification of MIDI ports is totally FUBAR'ed. this doesn't help
much.
:-) or mayby :-(
i'm not quite sure what you mean about ardour
(2's) behaviour, but its
not on my mind much at this point since ardour3 doesn't use ALSA MIDI
for anything at all, and will not do so again.
The names of input and output ports being identical. In this case
the port numbers are the same as well, so it's still possible to
look up a port by name and get the correct result.
Looking up by name is required if the port numbers can't be predicted.
It's also a bit more 'user friendly' to use names instead of numbers...
Regarding Ardour (A2) I do have another question.
It seems that sending the OSC command /ardour/transport_play
a second time has the effect of locating back to the previous
start position and resuming play from there. Could be useful,
but is this intentional ? The 'Start' widget does not behave
like that - a second click is just ignored.
Ciao,
--
FA
1:25 a.m.
On Fri, Apr 22, 2011 at 7:22 PM, Fons Adriaensen <fons(a)linuxaudio.org> wrote:
It seems that sending the OSC command
/ardour/transport_play
a second time has the effect of locating back to the previous
start position and resuming play from there. Could be useful,
but is this intentional ? The 'Start' widget does not behave
like that - a second click is just ignored.
its not really intentional. there is a way to click on the start
button that will do this (i don't recall what is is off the top of my
head, possibly button2 or some modifier), so its possible that the OSC
code is passing the wrong boolean to the function that does this
stuff.
1:05 a.m.
On Fri, Apr 22, 2011 at 10:30:55PM +0000, Fons Adriaensen wrote:
On Fri, Apr 22, 2011 at 06:06:54PM -0400, David
Robillard wrote:
I find this very strange as well.
I had a similar problem in my studio when I wanted to restore
connections with multiple ZynaddsubFX instances. All the port names
were the same, so I had no way of knowing which was which.
And the client ID is not much help either, because it completely
depends on the order in which you start the applications.
In the end, the only thing I could do was patch ZynaddsubFX so
it would make unique port names. It was very hard to code such
a simple thing, because like you, I couldn't find a way to get
an id from a name...
What's the use of having a name if it is not unique?
greetings,
lieven
Alsa sequencer names aren't unique or
intended for use as identifiers,
you probably don't want to do this if it's at all possible to avoid...
If that's true, what purpose do they have at all ?
In this concrete case, I have Python class which acts as
an interface to the Novation Launchpad. I want its ports
to connect to the launchpad, as any other connection
wouldn't make any sense at all. I know the names of
the required ports, but not the numbers.
Ciao,
--
FA
1:09 a.m.
On Fri, Apr 22, 2011 at 7:05 PM, Lieven Moors <lievenmoors(a)gmail.com> wrote:
I had a similar problem in my studio when I wanted to
restore
connections with multiple ZynaddsubFX instances. All the port names
were the same, so I had no way of knowing which was which.
And the client ID is not much help either, because it completely
depends on the order in which you start the applications.
In the end, the only thing I could do was patch ZynaddsubFX so
it would make unique port names. It was very hard to code such
a simple thing, because like you, I couldn't find a way to get
an id from a name...
why would you use the name at all. the way we handle this in ardour2
is to use only port IDs when storing MIDI connections. the names are
not used at all. we use pair<int,int> as SequencerPortAddress (the
first one is the client ID, the second is the port ID).
--p
1:19 a.m.
On Fri, Apr 22, 2011 at 07:09:24PM -0400, Paul Davis wrote:
On Fri, Apr 22, 2011 at 7:05 PM, Lieven Moors
<lievenmoors(a)gmail.com> wrote:
But surely, the client IDs are a moving target. I want to connect
to a specific instance of ZynaddsubFX, not to the program that was
the second one to get an ALSA ID...
Am I missing something?
lieven
I had a similar problem in my studio when I
wanted to restore
connections with multiple ZynaddsubFX instances. All the port names
were the same, so I had no way of knowing which was which.
And the client ID is not much help either, because it completely
depends on the order in which you start the applications.
In the end, the only thing I could do was patch ZynaddsubFX so
it would make unique port names. It was very hard to code such
a simple thing, because like you, I couldn't find a way to get
an id from a name...
why would you use the name at all. the way we handle this in ardour2
is to use only port IDs when storing MIDI connections. the names are
not used at all. we use pair<int,int> as SequencerPortAddress (the
first one is the client ID, the second is the port ID).
--p
1:23 a.m.
On Fri, Apr 22, 2011 at 7:18 PM, Lieven Moors <lievenmoors(a)gmail.com> wrote:
On Fri, Apr 22, 2011 at 07:09:24PM -0400, Paul Davis
wrote:
i don't think so. but since, as dave pointed out, there is no
particular mapping between client IDs and names, there's really no way
to do this. i think. like i said, i consider the whole design totally
FUBAR'ed. if you use numeric ID's you're screwed because the client ID
may change; if you use names, you're screwed because you can't map
reliably from names to the numeric ID's required by the API. that's
how it looks to me, anyway.
On Fri, Apr 22, 2011 at 7:05 PM, Lieven Moors
<lievenmoors(a)gmail.com> wrote:
But surely, the client IDs are a moving target. I want to connect
to a specific instance of ZynaddsubFX, not to the program that was
the second one to get an ALSA ID...
Am I missing something? I had a similar problem in my studio when I
wanted to restore
connections with multiple ZynaddsubFX instances. All the port names
were the same, so I had no way of knowing which was which.
And the client ID is not much help either, because it completely
depends on the order in which you start the applications.
In the end, the only thing I could do was patch ZynaddsubFX so
it would make unique port names. It was very hard to code such
a simple thing, because like you, I couldn't find a way to get
an id from a name...
why would you use the name at all. the way we handle this in ardour2
is to use only port IDs when storing MIDI connections. the names are
not used at all. we use pair<int,int> as SequencerPortAddress (the
first one is the client ID, the second is the port ID).
--p
1:43 a.m.
On Fri, Apr 22, 2011 at 07:22:59PM -0400, Paul Davis wrote:
i don't think so. but since, as dave pointed out,
there is no
particular mapping between client IDs and names, there's really no way
to do this. i think. like i said, i consider the whole design totally
FUBAR'ed. if you use numeric ID's you're screwed because the client ID
may change; if you use names, you're screwed because you can't map
reliably from names to the numeric ID's required by the API. that's
how it looks to me, anyway.
Things seem to be that way...
Regarding 'less than useful' interfaces, I just discovered
another one.
I want to use a Thinkpad's motion sensors for sound control.
You can read them from '/sys/devices/platform/hdasp/position',
which produces a string in the form (x,y). All good so far.
But opening that file and reading a line from it say every
50 ms just repeats the same values. To get updates you have
to close and re-open the file each time...
Ciao,
--
FA
11:43 a.m.
On Fri, 2011-04-22 at 19:22 -0400, Paul Davis wrote:
Well, hang on... can't you force a client to take a particular ID? You
can do that with the hardware addresses by specifying it in asound.conf
so your MIDI interfaces stay in the same place across reboots (otherwise
it's a race to see which interface initialises first).
Gordon MM0YEQ
But surely,
the client IDs are a moving target. I want to connect
to a specific instance of ZynaddsubFX, not to the program that was
the second one to get an ALSA ID...
Am I missing something?
i don't think so. but since, as dave pointed out, there is no
particular mapping between client IDs and names, there's really no way
to do this. i think. like i said, i consider the whole design totally
FUBAR'ed. if you use numeric ID's you're screwed because the client ID
may change; if you use names, you're screwed because you can't map
reliably from names to the numeric ID's required by the API. that's
how it looks to me, anyway. 
11:52 a.m.
Well, hang on... can't you force a client to take a particular ID? You
can do that with the hardware addresses by specifying it in asound.conf
so your MIDI interfaces stay in the same place across reboots (otherwise
it's a race to see which interface initialises first).
I managed to do that with different soundcards but I have not found a way
to
do that with the same soundcards (I have 3 amt8's I'd like to reliably map).
Greetz,
Immanuel
12:16 p.m.
On Sat, 2011-04-23 at 11:52 +0200, immanuel litzroth wrote:
Well, hang on... can't you force a client to take a particular
ID? You
can do that with the hardware addresses by specifying it in
asound.conf
so your MIDI interfaces stay in the same place across reboots
(otherwise
it's a race to see which interface initialises first).
I managed to do that with different soundcards but I have not found a
way to
do that with the same soundcards (I have 3 amt8's I'd like to reliably
map).
Greetz,
Immanuel
That's *probably* getting into udev territory - I haven't tried it.
Gordon MM0YEQ
1:55 a.m.
On Fri, 2011-04-22 at 22:30 +0000, Fons Adriaensen wrote:
On Fri, Apr 22, 2011 at 06:06:54PM -0400, David
Robillard wrote:
They are human-friendly /labels/, not /identifiers/.
Alsa sequencer names aren't unique or
intended for use as identifiers,
you probably don't want to do this if it's at all possible to avoid...
If that's true, what purpose do they have at all ? In this concrete case, I have Python class which acts
as
an interface to the Novation Launchpad. I want its ports
to connect to the launchpad, as any other connection
wouldn't make any sense at all. I know the names of
the required ports, but not the numbers.
Why do you know the names of the required ports but not the numbers?
-dr
P.S. The direct USB interface to the launchpad is pretty simple, I'm
driving mine without Alsa or MIDI anything at all
2:02 a.m.
On Fri, Apr 22, 2011 at 07:55:33PM -0400, David Robillard wrote:
Why do you know the names of the required ports but
not the numbers?
The names are lust 'Launchpad', the numbers are not fixed AFAIK.
P.S. The direct USB interface to the launchpad is
pretty simple, I'm
driving mine without Alsa or MIDI anything at all.
Yes, I know and using the USB interface was the original idea.
But I was surprised to see the thing appear in qjackctl when
plugged in, and decided to use the ALSA interface. But I'll
probably modify the code to use USB directly. As you say,
it's pretty simple.
Ciao,
--
FA
22 Apr
22 Apr
11:25 p.m.
On Friday 22 April 2011 22:45:34 Fons Adriaensen wrote:
after reading ASLA docs for some hours I've not
found an
answer to the following:
Given the ALSA sequencer client and port _names_, find the
numbers required to set up a connection (a 'subscription'
in ALSA lingo).
ALSA client and port names are not required to be unique.
I think the best you can do is query all clients and their ports
(snd_seq_query_next_client() etc.) until you find one with the name
you're looking for, and then just hope you got the right one...
By the way, aconnect seems to do the same (badly).
Dominic
11:55 p.m.
On Fri, Apr 22, 2011 at 11:24:47PM +0200, Dominic Sacré wrote:
On Friday 22 April 2011 22:45:34 Fons Adriaensen
wrote:
Aconnect seems to use 'snd_seq_parse_address()' which
is probably what I need. I found the docs by googling
that function's name, none of the docs I was able to
find otherwise mentioned it.
Badly indeed, since many clients just repeat the client
name as the port name, and even use the _same_ names for
input and output. Even Ardour does the latter. In this
case the corresponding port numbers are the same as well,
but without that it would be impossible to find them.
There's also the strange spaces that seem to be part of
the names:
client 129: 'ardour' [type=user]
0 'control '
1 'mcu '
2 'seq '
I've not yet found out if those spaces are required as
input to snd_seq_parse_address(). If they are, that's
bad since tools displaying the names (e.g. qjackctl)
don't show the spaces and a user would not have any
indication they part of the name.
Ciao,
--
FA
after reading ASLA docs for some hours I've
not found an
answer to the following:
Given the ALSA sequencer client and port _names_, find the
numbers required to set up a connection (a 'subscription'
in ALSA lingo).
ALSA client and port names are not required to be unique.
I think the best you can do is query all clients and their ports
(snd_seq_query_next_client() etc.) until you find one with the name
you're looking for, and then just hope you got the right one...
By the way, aconnect seems to do the same (badly).
4915
days inactive
4916
days old
linux-audio-dev@lists.linuxaudio.org
19 comments
7 participants
participants (7)
-
David Robillard -
Dominic Sacré -
Fons Adriaensen -
Gordon JC Pearce -
immanuel litzroth -
Lieven Moors -
Paul Davis