3 Nov
2020
3 Nov
'20
1:26 a.m.
Hi,
I try to understand why sysex go through jack MIDI sometimes, and
sometimes they are lost. I am aware that jack has a limit for the
maximum message size per period, but I am playing with program dump of
my Nordlead which are only 139 bytes -- definitely not huge.
I am running jackdbus with a2jmidid, through Catia. My system is Debian
+ kxstudio.
When I try an application with native jack MIDI, such as Reaper (with
jack backend) or MuSE, I can record sysex successfully, but when I play
it, it never reaches the hardware.
With applications using ALSA sequencer (tested with qtractor and
Rosegarden), I can record and play successfully sysex using ALSA.
However, I cannot record nor play sysex when connecting jack MIDI ports
instead (jack MIDI wrapper ports from a2jmidid).
Is this the expected behavior? Are we stuck with ALSA when we are
dealing with sysex messages?
Bonus question: is it possible to be warned when jack is dropping sysex
messages?
Thanks!
-a.
3 Nov
3 Nov
4:37 p.m.
Am 03.11.20 um 00:25 schrieb Alexandre DENIS:
When I try an application with native jack MIDI, such
as Reaper (with
jack backend) or MuSE, I can record sysex successfully, but when I play
it, it never reaches the hardware.
With applications using ALSA sequencer [...] I cannot record nor play sysex when
connecting jack MIDI ports
instead (jack MIDI wrapper ports from a2jmidid).
I made some tests and it seems that there is indeed a limit of 112 bytes
for outgoing sysex data.
Here's a very simple script using python-rtmidi I used to test (pass the
number of bytes to send as the first argument and an optional MIDI port
number as the second, if not, it will prompt for the one to use):
#!/usr/bin/env python
import sys
import time
import rtmidi
from rtmidi.midiutil import open_midioutput
port = sys.argv[2] if len(sys.argv) > 2 else None
midiout, name = open_midioutput(port, api=rtmidi.API_UNIX_JACK)
print("Opened port '%s'." % name)
data = bytearray(int(sys.argv[1]))
data[0] = 0xF0
data[-1] = 0xF7
print("Sending %d bytes" % len(data))
midiout.send_message(data)
time.sleep(0.5)
midiout.close_port()
del midiout
I have never encountered this limit before and I am not sure whether
this has always been so, or if it is a regression. I suggest reporting
this at https://github.com/jackaudio/jack2/issues
Chris
7:56 p.m.
On Tue, 3 Nov 2020, Christopher Arndt wrote:
I made some tests and it seems that there is indeed a
limit of 112 bytes
for outgoing sysex data.
I knew there was a limit but had never run into it. I realize that the
longest sysex I had used were ~20 bytes (mackie control with LCD info sent
channel at a time)
The one question I do have is the buffer size (128/2 ?) and if you tested
at more than one buffer size. I would expect that the two would be tied
together. I also remember reading that there had been code developed for
jack (I do not remember if it was jack 1 or 2 specific) to allow long,
non-realtime sysex to be used. However, I do not remember if it was ever
included or how it was accessed.
--
Len Ovens
www.ovenwerks.net
4 Nov
4 Nov
12:45 a.m.
On Tue, 3 Nov 2020 09:56:36 -0800 (PST)
Len Ovens <len(a)ovenwerks.net> wrote:
The one question I do have is the buffer size (128/2
?) and if you
tested at more than one buffer size. I would expect that the two
would be tied together.
Even with a buffer size of 2048, a sysex message of 139 bytes cannot be
sent (but can be received).
-a.
5 Nov
5 Nov
1:52 a.m.
On Tue, 3 Nov 2020 23:45:24 +0100
Alexandre DENIS <contact(a)alexandredenis.net> wrote:
On Tue, 3 Nov 2020 09:56:36 -0800 (PST)
Len Ovens <len(a)ovenwerks.net> wrote:
Somme follow-up.
I upgraded jackd from 1.9.12 to 1.9.16. It didn't change the result.
Then I tried to use the MIDI subsystem of the ALSA backend, instead of
a2jmidid. It works correctly. So the bug is actually in a2jmidid.
I had the habit to use a2jmidid, because I had previously a soundcard
supported only by FFADO, but now that my main soundcard is supported by
ALSA, I can switch.
However, what is annoying with MIDI through the ALSA backend is port
naming scheme. It doesn't follow ALSA naming scheme, so port grouping
is broken in Catia. It replaces spaces in names with dashes. And funny
thing, 'capture' and 'playback' are reversed in the aliases! Example,
from jack_lsp -A:
system:midi_playback_21
alsa_pcm:Prophet-12-Keyboard/midi_capture_1
Prophet-12-Keyboard:midi/capture_1
The main name here is *playback*, aliases are *capture*. WTF?
-a.
The one question I do have is the buffer size
(128/2 ?) and if you
tested at more than one buffer size. I would expect that the two
would be tied together.
Even with a buffer size of 2048, a sysex message of 139 bytes cannot
be sent (but can be received).
-a.
3:21 a.m.
from the jack(1)d man page:
-M, --midi-bufsize [ event-count ]
Specify the size of the buffer used for MIDI ports. Units are
"MIDI events per process() cycle", where an event occupies up to 4 bytes
(as most common MIDI events do). The default buffer size is 2048 bytes,
which is
about 500 MIDI events. Typical values for event-count will
range from 10 to about 1000. Be aware that using very high values along
with a large number of ports may cause JACK to fail to start because of
the amount
of memory that would be required.
On Tue, Nov 3, 2020 at 10:57 AM Len Ovens <len(a)ovenwerks.net> wrote:
On Tue, 3 Nov 2020, Christopher Arndt wrote:
I made some tests and it seems that there is
indeed a limit of 112 bytes
for outgoing sysex data.
I knew there was a limit but had never run into it. I realize that the
longest sysex I had used were ~20 bytes (mackie control with LCD info sent
channel at a time)
The one question I do have is the buffer size (128/2 ?) and if you tested
at more than one buffer size. I would expect that the two would be tied
together. I also remember reading that there had been code developed for
jack (I do not remember if it was jack 1 or 2 specific) to allow long,
non-realtime sysex to be used. However, I do not remember if it was ever
included or how it was accessed.
--
Len Ovens
www.ovenwerks.net
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
https://lists.linuxaudio.org/listinfo/linux-audio-user
4 Nov
4 Nov
6:31 a.m.
Tested MusE with Jack 1.9.14:
Successfully sent a 1536 byte sysex from one MusE instance to another.
Has Jack midi sysex length finally improved? Yay!
It sure didn't seem to be working years ago when I did the MusE sysex stuff.
Thanks.
Tim.
MusE sequencer project.
On 11/2/20 6:25 PM, Alexandre DENIS wrote:
Hi,
I try to understand why sysex go through jack MIDI sometimes, and
sometimes they are lost. I am aware that jack has a limit for the
maximum message size per period, but I am playing with program dump of
my Nordlead which are only 139 bytes -- definitely not huge.
I am running jackdbus with a2jmidid, through Catia. My system is Debian
+ kxstudio.
When I try an application with native jack MIDI, such as Reaper (with
jack backend) or MuSE, I can record sysex successfully, but when I play
it, it never reaches the hardware.
With applications using ALSA sequencer (tested with qtractor and
Rosegarden), I can record and play successfully sysex using ALSA.
However, I cannot record nor play sysex when connecting jack MIDI ports
instead (jack MIDI wrapper ports from a2jmidid).
Is this the expected behavior? Are we stuck with ALSA when we are
dealing with sysex messages?
Bonus question: is it possible to be warned when jack is dropping sysex
messages?
Thanks!
-a.
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
https://lists.linuxaudio.org/listinfo/linux-audio-user
6:36 a.m.
On 11/3/20 11:31 PM, Tim E. Real wrote:
Tested MusE with Jack 1.9.14:
Successfully sent a 1536 byte sysex from one MusE instance to another.
Has Jack midi sysex length finally improved? Yay!
It sure didn't seem to be working years ago when I did the MusE sysex
stuff.
Thanks.
Tim.
MusE sequencer project.
"Bonus question: is it possible to be warned when jack is dropping sysex
messages?"
I forgot to mention, in MusE we call jack_midi_event_reserve(),
�and if it fails, look for this message:
"MidiJackDevice::queueEvent ME_SYSEX: buffer overflow, sysex too big,
event lost"
I did not see the message in the test above. IIRC years ago it would happen.
Tim.
On 11/2/20 6:25 PM, Alexandre DENIS wrote:
> Hi,
>
> I try to understand why sysex go through jack MIDI sometimes, and
> sometimes they are lost. I am aware that jack has a limit for the
> maximum message size per period, but I am playing with program dump of
> my Nordlead which are only 139 bytes -- definitely not huge.
>
> I am running jackdbus with a2jmidid, through Catia. My system is Debian
> + kxstudio.
>
> When I try an application with native jack MIDI, such as Reaper (with
> jack backend) or MuSE, I can record sysex successfully, but when I play
> it, it never reaches the hardware.
>
> With applications using ALSA sequencer (tested with qtractor and
> Rosegarden), I can record and play successfully sysex using ALSA.
> However, I cannot record nor play sysex when connecting jack MIDI ports
> instead (jack MIDI wrapper ports from a2jmidid).
>
> Is this the expected behavior? Are we stuck with ALSA when we are
> dealing with sysex messages?
>
> Bonus question: is it possible to be warned when jack is dropping sysex
> messages?
>
> Thanks!
> -a.
>
>
> _______________________________________________
> Linux-audio-user mailing list
> Linux-audio-user(a)lists.linuxaudio.org
> https://lists.linuxaudio.org/listinfo/linux-audio-user
8:14 a.m.
On 11/3/20 11:36 PM, Tim E. Real wrote:
On 11/3/20 11:31 PM, Tim E. Real wrote:
And... sorry, a couple of more points:
The test above was at 128 jack buffer size.
Also, a slightly embarrassing but relevant gory detail about MusE:
If latency correction is enabled (see Settings), be aware of a caveat:
It works great! But an unfinished detail means that the first brief
�moment of audio or midi playback will NOT be heard.
This is because I will need to queue a small segment of playback
�BEFORE the transport starts. That's not done yet.
It means if you press play and there is a sysex or any other event
�placed right at that exact starting point, it will NOT be played.
Therefore, please place the red 'play head' cursor slightly BEFORE
�the desired start point, say about one quarter bar or more,
�before pressing play. All should be fine after that.
Thanks.
Tim.
Tested MusE with Jack 1.9.14:
Successfully sent a 1536 byte sysex from one MusE instance to another.
Has Jack midi sysex length finally improved? Yay!
It sure didn't seem to be working years ago when I did the MusE sysex
stuff.
Thanks.
Tim.
MusE sequencer project.
"Bonus question: is it possible to be warned when jack is dropping sysex
messages?"
I forgot to mention, in MusE we call jack_midi_event_reserve(),
�and if it fails, look for this message:
"MidiJackDevice::queueEvent ME_SYSEX: buffer overflow, sysex too big,
event lost"
I did not see the message in the test above. IIRC years ago it would
happen.
Tim.
On 11/2/20
6:25 PM, Alexandre DENIS wrote:
> Hi,
>
> I try to understand why sysex go through jack MIDI sometimes, and
> sometimes they are lost. I am aware that jack has a limit for the
> maximum message size per period, but I am playing with program dump of
> my Nordlead which are only 139 bytes -- definitely not huge.
>
> I am running jackdbus with a2jmidid, through Catia. My system is Debian
> + kxstudio.
>
> When I try an application with native jack MIDI, such as Reaper (with
> jack backend) or MuSE, I can record sysex successfully, but when I play
> it, it never reaches the hardware.
>
> With applications using ALSA sequencer (tested with qtractor and
> Rosegarden), I can record and play successfully sysex using ALSA.
> However, I cannot record nor play sysex when connecting jack MIDI ports
> instead (jack MIDI wrapper ports from a2jmidid).
>
> Is this the expected behavior? Are we stuck with ALSA when we are
> dealing with sysex messages?
>
> Bonus question: is it possible to be warned when jack is dropping sysex
> messages?
>
> Thanks!
> -a.
>
>
> _______________________________________________
> Linux-audio-user mailing list
> Linux-audio-user(a)lists.linuxaudio.org
> https://lists.linuxaudio.org/listinfo/linux-audio-user
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
https://lists.linuxaudio.org/listinfo/linux-audio-user 
9:29 a.m.
Just wondering, but maybe there are hardware limitations at the
receiving end (for hardware synths) that limit the size of a sysex? Or
the particular synth involved requires some additional setup to receive
sysexes?
I know my little Yamaha keyboard supports sysex of some sort but have
never bothered with it.
On 11/3/20 6:36 PM, Tim E. Real wrote:
On 11/3/20 11:31 PM, Tim E. Real wrote:
--
David W. Jones
gnome(a)hawaii.rr.com
authenticity, honesty, community
http://dancingtreefrog.com
"My password is the last 8 digits of π."
Tested MusE with Jack 1.9.14:
Successfully sent a 1536 byte sysex from one MusE instance to another.
Has Jack midi sysex length finally improved? Yay!
It sure didn't seem to be working years ago when I did the MusE sysex
stuff.
Thanks.
Tim.
MusE sequencer project.
"Bonus question: is it possible to be warned when jack is dropping sysex
messages?"
I forgot to mention, in MusE we call jack_midi_event_reserve(),
�and if it fails, look for this message:
"MidiJackDevice::queueEvent ME_SYSEX: buffer overflow, sysex too big,
event lost"
I did not see the message in the test above. IIRC years ago it would
happen.
Tim.
> On 11/2/20 6:25 PM, Alexandre DENIS wrote:
>> Hi,
>>
>> I try to understand why sysex go through jack MIDI sometimes, and
>> sometimes they are lost. I am aware that jack has a limit for the
>> maximum message size per period, but I am playing with program dump of
>> my Nordlead which are only 139 bytes -- definitely not huge.
>>
>> I am running jackdbus with a2jmidid, through Catia. My system is Debian
>> + kxstudio.
>>
>> When I try an application with native jack MIDI, such as Reaper (with
>> jack backend) or MuSE, I can record sysex successfully, but when I play
>> it, it never reaches the hardware.
>>
>> With applications using ALSA sequencer (tested with qtractor and
>> Rosegarden), I can record and play successfully sysex using ALSA.
>> However, I cannot record nor play sysex when connecting jack MIDI ports
>> instead (jack MIDI wrapper ports from a2jmidid).
>>
>> Is this the expected behavior? Are we stuck with ALSA when we are
>> dealing with sysex messages?
>>
>> Bonus question: is it possible to be warned when jack is dropping sysex
>> messages?
>>
>> Thanks!
>> -a. 
9:59 a.m.
On Tue, 3 Nov 2020 21:29:28 -1000
david <gnome(a)hawaii.rr.com> wrote:
Just wondering, but maybe there are hardware
limitations at the
receiving end (for hardware synths) that limit the size of a sysex? Or
the particular synth involved requires some additional setup to receive
sysexes?
I know my little Yamaha keyboard supports sysex of some sort but have
never bothered with it.
There are limits on some hardware synths.
Very many moons ago I wrote SYSEX software for an Acorn Archimedes to support a
couple of synths I had at the time. One of them transmitted SYSEX in a
series of quite small chunks, the other did the whole lot in a single stream.
Both comply with the MIDI standard, but you have to send back in the format you
receive, and as far as I recall there was no way of knowing this in advance.
This was with MIDI 0 so may have changed since then.
Both synths were Yamaha ones, and the biggie was "All voice and multi" -
basically the entire synth setup!
--
Will J Godfrey
http://www.musically.me.uk
Say you have a poem and I have a tune.
Exchange them and we can both have a poem, a tune, and a song.
6 Nov
6 Nov
7:18 a.m.
On 11/3/20 9:59 PM, Will Godfrey wrote:
On Tue, 3 Nov 2020 21:29:28 -1000
david <gnome(a)hawaii.rr.com> wrote:
All my hardware devices here are Yamaha. A couple of PSR models and an
S90 ES. Haven't done sysex stuff with any of them. I usually just use
them for their built-in voices or keyboard input. I suppose doing an
"All voice and multi" setup on the S90 would be a big sysex!
--
David W. Jones
gnome(a)hawaii.rr.com
authenticity, honesty, community
http://dancingtreefrog.com
"My password is the last 8 digits of π."
Just wondering, but maybe there are hardware
limitations at the
receiving end (for hardware synths) that limit the size of a sysex? Or
the particular synth involved requires some additional setup to receive
sysexes?
I know my little Yamaha keyboard supports sysex of some sort but have
never bothered with it.
There are limits on some hardware synths.
Very many moons ago I wrote SYSEX software for an Acorn Archimedes to support a
couple of synths I had at the time. One of them transmitted SYSEX in a
series of quite small chunks, the other did the whole lot in a single stream.
Both comply with the MIDI standard, but you have to send back in the format you
receive, and as far as I recall there was no way of knowing this in advance.
This was with MIDI 0 so may have changed since then.
Both synths were Yamaha ones, and the biggie was "All voice and multi" -
basically the entire synth setup!
1610
days inactive
1613
days old
linux-audio-user@lists.linuxaudio.org
11 comments
7 participants
participants (7)
-
Alexandre DENIS -
Christopher Arndt -
david -
Len Ovens -
Paul Davis -
Tim E. Real -
Will Godfrey