[linux-audio-dev] Re: [ardour-dev] xruns and recording

[Anthony]

* Paul Davis <paul at linuxaudiosystems.com> [Jul 22 03 18:44]:
> >I think there was some talk about xrun handling? What is the
> >behaviour of beta2? I'm seeing occasional xruns and then 
> >
> >** CRITICAL **: file gtk-canvas.c: line 1033 (gtk_canvas_item_hide): 
> >   assertion `item != NULL' failed.
> >
> >while recording. It's not clear to me if the xrun is causing the
> >recording to stop or if the xrun is a side effect of some bug.
> >
> >I've been trying to stress test things. 12 tracks simultaneous
> >recording 44.1Khz, 512*2 jack buffer. I've had the above eventually
> >happen everytime. CPU usage is < 15%, I wonder if its a disk 
> >issue. Switching desktops just completely killed ardour while
> >recording, normally this is not a problem:
> >
> >** CRITICAL **: file gtk-canvas.c: line 585
> >(gtk_canvas_item_set_valist): assert
> >ion `GTK_CANVAS_IS_CANVAS_ITEM (item)' failed. 
> 
> in gdb, do this:
> 
>    gdb> br main
>    gdb> run ...
>    .. stops at main ...
>    gdb> br g_logv
>    gdb> cont
> 
> then backtrace from there. its impossible to track these assertions
> otherwise, but very very easy to fix once you do. thanks.
> 
> --p

Of course I can't make them happen again. In that session I was doing
a lot to try and muck things up - changing zoom, recording overlapping
regions...that seems to be in line with canvas problems. I can still
make the recording dropout rather easily and receive and xrun, but that
seems to answer my question - its seems to be a performance issue
trigering something as opposed to the xrun being a result of a bug.
I'll try again later.

--ant