3 Jun
2010
3 Jun
'10
1:21 a.m.
I'm updating packages on a studio machine that's been fairly stable with
Jack 0.116.2, just to catch up with what's been going on. Jack's been
up to all kinds of stuff since then though.
There's Jack with DBus, Jack without DBus, Jack that does realtime by
default, Jack that needs '-R' for realtime, Jack specialized for
multiple processors, Jack that runs on multiple processors but isn't
specialized for it...there's probably a plain old Jack somewhere too.
I know Jack's been a busy guy, and all work and no play make Jack a dull
boy, but all these choices and options are enough to make someone go get
a bottle of Jack and get jacked up just thinking about it.
I'm running Gentoo with the Pro-Audio overlay, and it looks like there
are versions from both the 0.x and 1.x series available. What's
recommended these days for someone with four AMD cores, RME hardware
(via PCI-E, no firewire), and has always worked fine with the 0.x series
even on SMP? (And I have no PulseAudio, and don't plan on getting any.)
Sorry about the silly questions... As you can see, I don't know Jack.
--
+ Brent A. Busby + "We've all heard that a million monkeys
+ UNIX Systems Admin + banging on a million typewriters will
+ University of Chicago + eventually reproduce the entire works of
+ Physical Sciences Div. + Shakespeare. Now, thanks to the Internet,
+ James Franck Institute + we know this is not true." -Robert Wilensky
3 Jun
3 Jun
1:35 a.m.
Excerpts from Brent Busby's message of 2010-06-03 03:21:24 +0200:
I'm updating packages on a studio machine
that's been fairly stable with
Jack 0.116.2, just to catch up with what's been going on. Jack's been
up to all kinds of stuff since then though.
There's Jack with DBus, Jack without DBus, Jack that does realtime by
default, Jack that needs '-R' for realtime, Jack specialized for
multiple processors, Jack that runs on multiple processors but isn't
specialized for it...there's probably a plain old Jack somewhere too.
I know Jack's been a busy guy, and all work and no play make Jack a dull
boy, but all these choices and options are enough to make someone go get
a bottle of Jack and get jacked up just thinking about it.
I'm running Gentoo with the Pro-Audio overlay, and it looks like there
are versions from both the 0.x and 1.x series available. What's
recommended these days for someone with four AMD cores, RME hardware
(via PCI-E, no firewire), and has always worked fine with the 0.x series
even on SMP? (And I have no PulseAudio, and don't plan on getting any.)
Sorry about the silly questions... As you can see, I don't know Jack.
Hi Brent.
I'll try to keep it simple:
- 0.x is jack1, plain old jack. It works on any machine but only on a
single core.
- 1.x is jack2, just another implementation. It works on any machine and
takes advantage of smp
There are some implementations that I just call 'minor' here, and every
implementation has its pros and cons. If you're happy with jack1 you can
stick with it, it's the earliest implementation afaik and hence probably
the most stable one.
--
Regards,
Philipp
--
"Wir stehen selbst enttäuscht und sehn betroffen / Den Vorhang zu und alle Fragen
offen." Bertolt Brecht, Der gute Mensch von Sezuan
9:49 a.m.
Brent Busby wrote:
I'm updating packages on a studio machine
that's been fairly stable with
Jack 0.116.2, just to catch up with what's been going on. Jack's been
up to all kinds of stuff since then though.
There's Jack with DBus, Jack without DBus, Jack that does realtime by
default, Jack that needs '-R' for realtime, Jack specialized for
multiple processors, Jack that runs on multiple processors but isn't
specialized for it...there's probably a plain old Jack somewhere too.
I know Jack's been a busy guy, and all work and no play make Jack a dull
boy, but all these choices and options are enough to make someone go get
a bottle of Jack and get jacked up just thinking about it.
I'm running Gentoo with the Pro-Audio overlay, and it looks like there
are versions from both the 0.x and 1.x series available. What's
recommended these days for someone with four AMD cores, RME hardware
(via PCI-E, no firewire), and has always worked fine with the 0.x series
even on SMP? (And I have no PulseAudio, and don't plan on getting any.)
I don't know all those details. JackDMP 1.9.x works here where I
couldn't get JACK 0.116 to work. But I've only got lowly single-core
processors here.
--
David
gnome(a)hawaii.rr.com
authenticity, honesty, community
10:43 a.m.
On Wed, 2 Jun 2010, david wrote:
I don't know all those details. JackDMP 1.9.x
works here where I
couldn't get JACK 0.116 to work. But I've only got lowly single-core
processors here.
Hmm... That's mostly what I was wanting to know. I've heard the 0.x
series is more mature, and I've been using it -- but does that mean the
1.x series is deprecated? Is it working for most people?
--
+ Brent A. Busby + "We've all heard that a million monkeys
+ UNIX Systems Admin + banging on a million typewriters will
+ University of Chicago + eventually reproduce the entire works of
+ Physical Sciences Div. + Shakespeare. Now, thanks to the Internet,
+ James Franck Institute + we know this is not true." -Robert Wilensky
10:50 a.m.
Excerpts from Brent Busby's message of 2010-06-03 12:43:39 +0200:
On Wed, 2 Jun 2010, david wrote:
Neither is the successor of the other, neither is deprecated. It's two
implementations developed in parallel.
--
Regards,
Philipp
--
"Wir stehen selbst enttäuscht und sehn betroffen / Den Vorhang zu und alle Fragen
offen." Bertolt Brecht, Der gute Mensch von Sezuan
I don't know all those details. JackDMP 1.9.x
works here where I
couldn't get JACK 0.116 to work. But I've only got lowly single-core
processors here.
Hmm... That's mostly what I was wanting to know. I've heard the 0.x
series is more mature, and I've been using it -- but does that mean the
1.x series is deprecated? Is it working for most people? 
9:02 p.m.
On Thu, Jun 03, 2010 at 11:47:21AM +0200, Philipp ??berbacher wrote:
Excerpts from Brent Busby's message of 2010-06-03
12:43:39 +0200:
Wait, what?
Last I was following this, Paul was saying fairly consistently that jackdmp was intended
to replace jackd. Newer codebase, better implementation, C++ not C, multi-core, etc etc.
Has that position changed while I wasn't paying attention?
-ken
On Wed, 2 Jun 2010, david wrote:
Neither is the successor of the other, neither is deprecated. It's two
implementations developed in parallel. I don't know all those details. JackDMP 1.9.x
works here where I
couldn't get JACK 0.116 to work. But I've only got lowly single-core
processors here.
Hmm... That's mostly what I was wanting to know. I've heard the 0.x
series is more mature, and I've been using it -- but does that mean the
1.x series is deprecated? Is it working for most people? 
9:07 p.m.
On Thu, 3 Jun 2010, Ken Restivo wrote:
Yes. I think it hit the fan around the beginning of April
when someone (Adi?) announced that many of the Linux
distro's are doing a coordinated switch to Jack2.
At that time, the position was changed.
-gabriel
Neither is the
successor of the other, neither is deprecated. It's two
implementations developed in parallel.
Wait, what?
Last I was following this, Paul was saying fairly
consistently that jackdmp was intended to replace jackd.
Newer codebase, better implementation, C++ not C,
multi-core, etc etc.
Has that position changed while I wasn't paying attention? 
9:16 p.m.
On Thu, Jun 03, 2010 at 02:00:11PM -0700, Ken Restivo wrote:
Last I was following this, Paul was saying fairly
consistently that jackdmp was intended to replace jackd. Newer codebase, better
implementation, C++ not C, multi-core, etc etc.
Has that position changed while I wasn't paying attention?
Apparently it has, AFAICS the result of
- personal preferences or allergies of some of the jack1
developers who would probably drop out if jack1 is
declared 'dead',
- and Sletz and the jack2 team not being too eager to
take it all on their shoulders alone.
Ciao,
--
FA
O tu, che porte, correndo si ?
E guerra e morte !
12:15 p.m.
On Thu, Jun 3, 2010 at 12:52 PM, James Warden <warjamy(a)yahoo.com> wrote:
1.9.6 is working great on my dual core system, in synchronous mode (jackd
-S).
As regards jack master transport, latest ardour2 is working fine with jackd
0.118.0 but it still has this little problem with jackdmp 1.9.5:
http://tracker.ardour.org/bug_view_advanced_page.php?bug_id=2856
Will this be resolved with jack 1.9.6?
Regards, Pablo
4:56 p.m.
Quoting Pablo Fernandez <pablo.fbus(a)gmail.com>om>:
As regards jack master transport, latest ardour2 is
working fine with jackd
0.118.0 but it still has this little problem with jackdmp 1.9.5:
http://tracker.ardour.org/bug_view_advanced_page.php?bug_id=2856
When you run 1.9.5 (with Ardour 2.8.8 and Hydrogen 0.9.5-beta2 at
least) you don't have to set Ardour as Master.
Without it, if I click anywhere on the timeline in Ardour, Hydrogen
jumps to that spot.
It also works the other way around if I click somewhere in Hydrogen.
10:44 p.m.
As regards jack master transport, latest ardour2 is working fine with
Yes, but then, who is the tempo master? I have an ardour song with several
tempo changes I made from a tempo map with klick2ardour. I have hydrogen
which follows the tempo changes perfectly but only as long as I set ardour
transport as jack master. I have just tested it again with jack2. I am back
to jack1.
Regards, Pablo
jackd
0.118.0 but it still has this little problem with jackdmp 1.9.5:
http://tracker.ardour.org/bug_view_advanced_page.php?bug_id=2856
When you run 1.9.5 (with Ardour 2.8.8 and Hydrogen 0.9.5-beta2 at least)
you don't have to set Ardour as Master.
Without it, if I click anywhere on the timeline in Ardour, Hydrogen jumps
to that spot.
It also works the other way around if I click somewhere in Hydrogen.
9:01 p.m.
On Thu, Jun 03, 2010 at 03:52:11AM -0700, James Warden wrote:
What is synchronous mode? (A link to the place in the docs would be a plenty good answer,
sorry, I don't feel like fishing around in it right now).
-ken
Hmm...? That's mostly what I was wanting to
know.? I've heard the 0.x series is more mature, and
I've been using it -- but does that mean the 1.x series is
deprecated?? Is it working for most people?
1.9.6 is working great on my dual core system, in synchronous mode (jackd -S).
9:07 p.m.
On Thu, Jun 03, 2010 at 01:58:49PM -0700, Ken Restivo wrote:
What is synchronous mode? (A link to the place in the
docs would be a plenty good answer, sorry, I don't feel like fishing around in it
right now).
AFAIK, synchronous mode is what jack1 does:
read from soundcard
process graph
write to soundcard
while jack2's default mode is
write output of previous cycle to soundcard
read from soundcard
process graph
which is somewhat more resilient to occasional xruns,
but adds one cycle of latency.
Ciao,
--
FA
O tu, che porte, correndo si ?
E guerra e morte !
9:51 p.m.
Excerpts from fons's message of 2010-06-03 23:02:46 +0200:
On Thu, Jun 03, 2010 at 01:58:49PM -0700, Ken Restivo
wrote:
According to http://www.grame.fr/~letz/Timing.pdf asynchronous mode is
only sensible to use when your audio cards driver interrupt is regular.
It also more or less explains how this can be measured.
I'm not totally sure how to interpret it properly in my case, but it's
a USB audio interface, which might be problematic by definition.
http://murks.lima-city.de/serendipity/uploads/timing1.png
In a sense it's regular... +/- 500usec all the way...
Well, smp doesn't make sense on my machine anyway, so I stick with jack1
(well, tschack atm.).
--
Regards,
Philipp
--
"Wir stehen selbst enttäuscht und sehn betroffen / Den Vorhang zu und alle Fragen
offen." Bertolt Brecht, Der gute Mensch von Sezuan
What is synchronous mode? (A link to the place in
the docs would be a plenty good answer, sorry, I don't feel like fishing around in it
right now).
AFAIK, synchronous mode is what jack1 does:
read from soundcard
process graph
write to soundcard
while jack2's default mode is
write output of previous cycle to soundcard
read from soundcard
process graph
which is somewhat more resilient to occasional xruns,
but adds one cycle of latency.
Ciao,
4 Jun
4 Jun
4:47 a.m.
On Thu, Jun 03, 2010 at 05:02:22PM -0500, Gabriel M. Beddingfield wrote:
On Thu, 3 Jun 2010, Philipp ??berbacher wrote:
Wait wait wait, there's now a THIRD jackd, a fork of jack 1 codebase with SMB
support?
Jeez, it has gotten complicated hasn't it?
-ken
Well, smp doesn't make sense on my machine
anyway, so I stick with jack1
(well, tschack atm.).
??? tschack is smp, isn't it?
10:53 a.m.
On Fri, Jun 4, 2010 at 12:45 AM, Ken Restivo <ken(a)restivo.org> wrote:
Wait wait wait, there's now a THIRD jackd, a fork
of jack 1 codebase with SMB support?
Jeez, it has gotten complicated hasn't it?
which DNS server do you use? firefox, mozilla, iceweasel? emacs,
xemacs, lucid emacs? oh, and of course, ubuntu, debian, fedora, arch,
gentoo, suse, and others too numerous to mention ...
11:40 a.m.
On Fri, Jun 4, 2010 at 12:45 AM, Ken Restivo
<ken(a)restivo.org> wrote:
The beauty of open source development.
--
Patrick Shirkey
Boost Hardware Ltd.
Wait wait wait, there's now a THIRD jackd, a
fork of jack 1 codebase
with SMB support?
Jeez, it has gotten complicated hasn't it?
which DNS server do you use? firefox, mozilla, iceweasel? emacs,
xemacs, lucid emacs? oh, and of course, ubuntu, debian, fedora, arch,
gentoo, suse, and others too numerous to mention ...
10:09 p.m.
On 4 June 2010 19:40, <pshirkey(a)boosthardware.com> wrote:
Dang. Nothing to counter with :(
--
GPG/PGP ID: B42DDCAD
On Fri, Jun 4,
2010 at 12:45 AM, Ken Restivo <ken(a)restivo.org> wrote:
The beauty of open source development. Wait wait wait, there's now a THIRD jackd, a
fork of jack 1 codebase
with SMB support?
Jeez, it has gotten complicated hasn't it?
which DNS server do you use? firefox, mozilla, iceweasel? emacs,
xemacs, lucid emacs? oh, and of course, ubuntu, debian, fedora, arch,
gentoo, suse, and others too numerous to mention ...
6 Jun
6 Jun
9:51 a.m.
On Fri, Jun 4, 2010 at 5:53 AM, Paul Davis <paul(a)linuxaudiosystems.com> wrote:
which DNS server do you use? firefox, mozilla,
iceweasel? emacs,
xemacs, lucid emacs? oh, and of course, ubuntu, debian, fedora, arch,
gentoo, suse, and others too numerous to mention ...
yes, but firefox, mozilla, iceweasel, emacs, xemacs, lucid emacs, etc.
can happily coexist on my system with no problem. in fact, I can fire
all of them up all at once.
various flavors of jack on the other hand...
--
Josh Lawrence
http://www.hardbop200.com
11:05 a.m.
On Sun, 2010-06-06 at 04:51 -0500, Josh Lawrence wrote:
On Fri, Jun 4, 2010 at 5:53 AM, Paul Davis
<paul(a)linuxaudiosystems.com> wrote:
Can coexist on your system, and even be fired all at once (if you have
different audio interfaces for the different flavors). I'm pretty sure
that causes xruns though.
which DNS server do you use? firefox, mozilla,
iceweasel? emacs,
xemacs, lucid emacs? oh, and of course, ubuntu, debian, fedora, arch,
gentoo, suse, and others too numerous to mention ...
yes, but firefox, mozilla, iceweasel, emacs, xemacs, lucid emacs, etc.
can happily coexist on my system with no problem. in fact, I can fire
all of them up all at once.
various flavors of jack on the other hand...
11:24 a.m.
On 06/06/2010 11:51 AM, Josh Lawrence wrote:
On Fri, Jun 4, 2010 at 5:53 AM, Paul Davis
<paul(a)linuxaudiosystems.com> wrote:
well, try to make sendmail, exim, and postfix coexist peacefully on your
machine. with that experience up your sleeve, i'm sure contributing
patches to pull the same trick with three jack daemons is matter of
minutes :)
which DNS server do you use? firefox, mozilla,
iceweasel? emacs,
xemacs, lucid emacs? oh, and of course, ubuntu, debian, fedora, arch,
gentoo, suse, and others too numerous to mention ...
yes, but firefox, mozilla, iceweasel, emacs, xemacs, lucid emacs, etc.
can happily coexist on my system with no problem. in fact, I can fire
all of them up all at once.
various flavors of jack on the other hand... 
2:58 p.m.
Hi,
On Sunday 06 June 2010 14:09:45 drew Roberts wrote:
On Sunday 06 June 2010 07:24:27 Jörn Nettingsmeier
wrote:
Then have them all use different mail spools. And different users. But from the
same user-database...
Running several jackd in parallel is easier when you use different devices and
identify them by special names. The problems arise from apps that don't allow
you to choose the jackd to connect to...
Have fun,
Arnold
well, try to make sendmail, exim, and postfix
coexist peacefully on your
machine.
Hmm, my first thought would be to put in 3 nics and make each listen to
only one nic. 
8:39 p.m.
On 06/06/2010 04:58 PM, Arnold Krille wrote:
Hi,
On Sunday 06 June 2010 14:09:45 drew Roberts wrote:
> On Sunday 06 June 2010 07:24:27 Jörn Nettingsmeier wrote:
>> well, try to make sendmail, exim, and postfix coexist peacefully on your
>> machine.
>
> Hmm, my first thought would be to put in 3 nics and make each listen to
> only one nic.
..and /usr/sbin/sendmail would do what?
Then have them all use different mail spools. And
different users. But from the
same user-database...
Running several jackd in parallel is easier when you use different devices and
identify them by special names. The problems arise from apps that don't allow
you to choose the jackd to connect to...
Not really. One can always use the $JACK_DEFAULT_SERVER environment
variable to override the 'default' server.
It's documented at
http://jackaudio.org/files/docs/html/group__ClientFunctions.html#ga28977ad0…
2:32 p.m.
On Sun, Jun 6, 2010 at 4:51 AM, Josh Lawrence <hardbop200(a)gmail.com> wrote:
yes, but firefox, mozilla, iceweasel, emacs, xemacs,
lucid emacs, etc.
can happily coexist on my system with no problem. in fact, I can fire
all of them up all at once.
try that with different versions of an NFS server ....
various flavors of jack on the other hand...
jack is a server plus a library. if the versions don't match, kaboom.
the other examples are either applications or just libraries. there's
not a whole lot of *nix software that consists of a server/library
pair. X11 would be the other canonical example.
5:21 p.m.
On Sun, Jun 06, 2010 at 09:31:59AM -0500, Paul Davis wrote:
On Sun, Jun 6, 2010 at 4:51 AM, Josh Lawrence
<hardbop200(a)gmail.com> wrote:
Hmm. And I guess postfix/sendmail/exim4/ssmtp would be another, so I see your point.
However, with my distro, I can trivially easily switch back and forth between
MTA/MUA's, in fact, I just switched between a bunch of them that last week.
Jack/Ardour's allergy to distributions in general (and Debian in particular) has made
this harder to manage. If I could go "sudo apt-get install
jack-flavor-of-the-month" and it switches over, then go "sudo apt-get install
jackd-original" to switch back, then it's no problemo.
There are also other servers/services like Apache/Nginx/thttpd/lighthttpd, which *can*
coexist or even all run simultaneously on a single machine. That might be another model
for making the various jacks easier to switch between. I remember some discussion of this
many months or years ago, but I don't know how that turned out.
-ken
yes, but firefox, mozilla, iceweasel, emacs,
xemacs, lucid emacs, etc.
can happily coexist on my system with no problem. ?in fact, I can fire
all of them up all at once.
try that with different versions of an NFS server ....
various flavors of jack on the other hand...
jack is a server plus a library. if the versions don't match, kaboom.
the other examples are either applications or just libraries. there's
not a whole lot of *nix software that consists of a server/library
pair. X11 would be the other canonical example. 
6:07 p.m.
Just occurred to me - this subject could be a song title.
OK,OK, I'll pick up my coat on the way out :)
--
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 Jun
6 Jun
6:17 p.m.
On Sun, 6 Jun 2010, Ken Restivo wrote:
Jack/Ardour's allergy to distributions in general
(and
Debian in particular) has made this harder to manage. If I
could go "sudo apt-get install jack-flavor-of-the-month"
and it switches over, then go "sudo apt-get install
jackd-original" to switch back, then it's no problemo.
Jack+Debian: This situation will be improving. Probably
won't be in time for squeeze, though. Not a simple problem
to fix, though... because applications have to link to
/something/ -- and the library follows the jackd
implementation.
-gabriel
7:55 p.m.
On Sun, Jun 06, 2010 at 01:17:37PM -0500, Gabriel M. Beddingfield wrote:
On Sun, 6 Jun 2010, Ken Restivo wrote:
Wait, what? All the apps have to be recompiled in order to switch jack implementations?
If that's the case, then having multiple jack implementations out there is just short
of insane.
-ken
Jack/Ardour's allergy to distributions in
general (and Debian in
particular) has made this harder to manage. If I could go "sudo apt-get
install jack-flavor-of-the-month" and it switches over, then go "sudo
apt-get install jackd-original" to switch back, then it's no problemo.
Jack+Debian: This situation will be improving. Probably won't be in
time for squeeze, though. Not a simple problem to fix, though... because
applications have to link to /something/ -- and the library follows the
jackd implementation.
8:05 p.m.
Excerpts from Ken Restivo's message of 2010-06-06 21:53:10 +0200:
On Sun, Jun 06, 2010 at 01:17:37PM -0500, Gabriel M.
Beddingfield wrote:
I switched between tschack, jack1 and jack2 without problems. Not sure
there are potential problems, but I didn't experience any.
--
Regards,
Philipp
--
"Wir stehen selbst enttäuscht und sehn betroffen / Den Vorhang zu und alle Fragen
offen." Bertolt Brecht, Der gute Mensch von Sezuan
On Sun, 6 Jun 2010, Ken Restivo wrote:
Wait, what? All the apps have to be recompiled in order to switch jack implementations?
If that's the case, then having multiple jack implementations out there is just short
of insane.
-ken Jack/Ardour's allergy to distributions in
general (and Debian in
particular) has made this harder to manage. If I could go "sudo apt-get
install jack-flavor-of-the-month" and it switches over, then go "sudo
apt-get install jackd-original" to switch back, then it's no problemo.
Jack+Debian: This situation will be improving. Probably won't be in
time for squeeze, though. Not a simple problem to fix, though... because
applications have to link to /something/ -- and the library follows the
jackd implementation.
8:11 p.m.
On Sun, 6 Jun 2010, Ken Restivo wrote:
No, they don't have to be recompiled.
Jack+Debian:
This situation will be improving. Probably won't be in
time for squeeze, though. Not a simple problem to fix, though... because
applications have to link to /something/ -- and the library follows the
jackd implementation.
Wait, what? All the apps have to be recompiled in order to
switch jack implementations? From a packaging point-of-view, that's what makes
it so
hard. I'm not at all an expert on the subject, but the
confusion goes something like this:
Package `cool_tracker` requires `libjack0`
`libjack0` is supplied by Jack1 _and_ Jack2. If you switch
jackd's, you have to switch libjack0. So, then you get
something like:
libjack0-jack1 (provides libjack0)
libjack0-jack2 (provides libjack0)
Suppose user does:
$ apt-get cool_tracker
Now, which one gets installed by default?
And when a someone packages `cool_tracker`
$ gcc -o cool_tracker `pkg-config --cflags --libs jack` cool_tracker.c
Then cool_tracker has a link to /usr/lib/libjack.so.0
So... you have to make sure that the .deb package says it
needs the virtual libjack0 package instead of libjack0-jack2
(which is what the packager has installed).
THEN, you run into ABI issues between Jack1 and Jack2.
Theoretically, there should be none -- but last I
check, Jack2 is missing some of Jack1's symbols.
-gabriel
9:41 p.m.
On Sun, Jun 06, 2010 at 03:11:05PM -0500, Gabriel M. Beddingfield wrote:
On Sun, 6 Jun 2010, Ken Restivo wrote:
No, they don't have to be recompiled.
From a packaging point-of-view, that's what makes it so hard. I'm not at
all an expert on the subject, but the confusion goes something like this:
Package `cool_tracker` requires `libjack0`
`libjack0` is supplied by Jack1 _and_ Jack2. If you switch jackd's, you
have to switch libjack0. So, then you get something like:
libjack0-jack1 (provides libjack0)
libjack0-jack2 (provides libjack0)
Suppose user does:
$ apt-get cool_tracker
Now, which one gets installed by default?
And when a someone packages `cool_tracker`
$ gcc -o cool_tracker `pkg-config --cflags --libs jack` cool_tracker.c
Then cool_tracker has a link to /usr/lib/libjack.so.0
So... you have to make sure that the .deb package says it needs the
virtual libjack0 package instead of libjack0-jack2 (which is what the
packager has installed).
THEN, you run into ABI issues between Jack1 and Jack2. Theoretically,
there should be none -- but last I check, Jack2 is missing some of
Jack1's symbols.
OK thanks. My apologies for making you rehash stuff that has probably been flying around
on the jack list for a while.
Still, the average, non-jack-developers-mailing-list-following linux audio user should be
spared from having to know all this stuff, and hopefully the various jack-flavor
maintainers and distro maintainers are all working together to make this usable.
Still, if ABI compatibility isn't there, this feels kind of dangerous and could get
very ugly.
I now see why so many people reflexively advise anyone having any trouble with linux audio
to just get a livecd of a different distro and try that instead. If ABI's aren't
compatible between the different jack's, then task of setting up linux audio with jack
will fall out of the hands of most users and onto those who maintain custom,
mutually-incompatible distros instead. Possibly not a bad thing, and we might be there
already as people seek out pre-compiled RT kernels, but perhaps pushing linux audio even
farther into custom purpose-built distros might have unintended consequences.
-ken
Jack+Debian: This situation will be improving. Probably won't be in
time for squeeze, though. Not a simple problem to fix, though... because
applications have to link to /something/ -- and the library follows the
jackd implementation.
Wait, what? All the apps have to be recompiled in order to switch jack
implementations?
10:07 p.m.
Before I'm accused of slinging FUD...
Still, if ABI compatibility isn't there, this
feels kind
of dangerous and could get very ugly.
The ABI compatablity issue came up with a rarely-used
feature of the Jack API.[1] Jack2 actually /does/ include
it, but for some reason it got lost when linked. It was
never noticed until Debian switch to Jack2 and found two
applications break.
It's really minor.
-gabriel
[1] I think it was the global variable jack_error_callback.
http://jackaudio.org/files/docs/html/group__ErrorOutput.html#ga2f71410a27ce…
7 Jun
7 Jun
4:17 a.m.
On Sun, Jun 6, 2010 at 5:38 PM, Ken Restivo <ken(a)restivo.org> wrote:
If ABI's aren't compatible between the
different jack's, then task of setting up linux audio with jack will fall out of the
hands of most users and onto those who maintain custom, mutually-incompatible distros
instead.
if different versions of JACK do not share an ABI (at least, the core
API) then there is a mistake on the part of JACK developers and/or
packagers (either all or some of us). it is absolutely not part of the
design, the plan, the intent or whatever. it is a mistake to ever
conclude that an ABI incompatibility between versions is "how it
goes". if it exists, its a bug and its needs to be fixed, ASAP.
7:39 a.m.
Paul Davis wrote:
On Sun, Jun 6, 2010 at 5:38 PM, Ken Restivo
<ken(a)restivo.org> wrote:
The debian-multimedia effort to put both into debian will help find these bugs,
and probably help to solve them too. Then having both interchangeable in debian
should make sure any incompatibilities that creep in get picked up fairly
quickly. Probably it will be after Squeeze, but it should be in Sid fairly soon
by the sounds of the mailing list, it seemed like quite an effort to get this far.
Simon
If ABI's aren't compatible between the
different jack's, then task of setting up linux audio with jack will fall out of the
hands of most users and onto those who maintain custom, mutually-incompatible distros
instead.
if different versions of JACK do not share an ABI (at least, the core
API) then there is a mistake on the part of JACK developers and/or
packagers (either all or some of us). it is absolutely not part of the
design, the plan, the intent or whatever. it is a mistake to ever
conclude that an ABI incompatibility between versions is "how it
goes". if it exists, its a bug and its needs to be fixed, ASAP.
5104
days inactive
5108
days old
linux-audio-user@lists.linuxaudio.org
38 comments
20 participants
participants (20)
-
Arnold Krille -
Brent Busby -
david -
drew Roberts -
Folderol -
fons＠kokkinizita.net -
Gabriel M. Beddingfield -
James Warden -
Josh Lawrence -
Jörn Nettingsmeier -
Ken Restivo -
Ng Oon-Ee -
Pablo Fernandez -
Paul Davis -
Peder Hedlund -
Philipp Überbacher -
pshirkey＠boosthardware.com -
Ray Rashif -
Robin Gareus -
Simon Wise