Bug report for @brad: Upgrading to build 4197 or 4198 (Cantabile Performer x64) causes some of my bindings to go into infinite loops when triggered, freezing Cantabile and forcing me to task-kill it to escape. I’ve downgraded back to 4196 for now.
After adding this binding to an empty song, pressing controller button 87 once causes the Transpose to increment in an infinite loop (until it reaches +127 and can’t go any farther, but it keeps trying). Cantabile freezes and I have to kill the process to regain control. The same binding on 4196 or lower doesn’t exhibit this behavior.
That fixes the binding loop, but I’m still seeing some kind of timing instability in 4197-4199.
Infrequently (e.g., once or twice during a 2-hour rehearsal) Cantabile will suddenly “stutter” for several seconds, then freeze for several more seconds, then sound like the fast-forward button has been pressed on an old-fashioned tape deck, and then finally resume normal operation. The “stuttering” sounds like a machine gun effect that replays a slice of about 100ms of whatever sound was playing at the time of the interruption.
Unfortunately I can’t find any reliable way to reproduce the problem for you. It seems to happen at completely random times, and I usually have dozens of plugins running, making it hard to isolate. And there’s no crash file since Cantabile doesn’t crash. Maybe some part of the new video sync code has destabilized Cantabile’s timing? In case it helps, I only ever use Cantabile with the metronome set as the master transport, and I don’t have any externally synced devices.
Yes, I’ve been watching that thread with interest along with this one. But this felt like it might be a different bug, so I thought I’d better report it just in case.
Unfortunately my fractional tempos are broken, which you quickly fixed in 4199 (thank you!), so for now I’m kinda stuck between builds.
(Posting this here since the new topic is now closed.)
I just experienced the above bug again on v4202 (unofficial build). I still have no idea what triggers it, though. I couldn’t get it to happen on v4201 irrespective of whether the “slow” box was checked, but it happens so rarely that maybe I just didn’t use that version for long enough.
But this felt like it might be a different bug, so I thought I’d better report it just in case.