On Mon, Oct 19, 2015 at 4:55 PM, Ivica Ico Bukvic <ico@vt.edu> wrote:

On 10/18/2015 10:31 AM, Paul Davis wrote:

On Sun, Oct 18, 2015 at 10:23 AM, Len Ovens <len@ovenwerks.net> wrote:

USB hot plug _can_ work with jack:
http://gareus.org/blog/jack2dbus

right. if the software notices signals on D-bus that the device is going away, then it stops using it in a way that causes ALSA to get locked up, putting the process into an un-killable state.

it would be nice if ALSA could be fixed to not do this, but it would also be nice if key software like JACK would pay attention to device state information too.

Since former is not necessarily in our immediate control, any hope of latter being implemented by those on this list who are in control?

well, jack2 will listen to dbus already (see the link to robin's work on this). jack1 is never going to interact directly with d-bus in the same way as jack2, but for this particular purpose, a platform centric patch would be smart.

better yet though, this should be buried inside the ALSA backend and all backends should have a way to tell JACK that the device has "gone away".we've had to visit+revisit this for the new abstracted audio I/O backends inside ardour already.

chances of any of the usual suspects doing this for JACK1: close to zero i think.