Sure, there is a topic on GitHub you can add things to. It might be a little hard for us to do anything without a reproducer.
If you think it might be GC related you at try sprinkling this though out the code. 10000.do{ 0!(4192.rand) }
There’s clearly either a bug in the Qt collider code, or in the link between them.
The GC one, I think it’s because the keydown responder is throwing an error (probably because the list is empty) and then it crashes when building the call stack.
The pure Qt bug, I don’t know what’s up with that. 999 times out of 1000, the ListView behaves normally; once in a great while, it crashes (but never when the interpreter load is light).
I know it’s hard without a reproducer, but it’s a bad bug. Without a fix I’d have to stay on 3.14 for shows (I know, 3.15 is not ready for that kind of prime time, I’m just kicking the tires at home).
hjh
Yeah this should be fixed by reference counting the frames.
The next dev meeting will be this upcoming Wednesday @ 2026-10-14T18:00:00Z→2026-10-14T20:00:00Z via https://video.imm.rsh-duesseldorf.de/sc-dev-meeting.
Everyone is welcome - people who want to start contributing, people who stopped contributing, people who just want to lurk 
We already have some topics collected via Dev meeting · GitHub ,but feel free to also bring your topics/PRs/questions into the meeting!
We will also discuss the road towards 3.15.
The CoC applies.
Looking forward to see you!
PS: Feel free to also use the forum chat @ Chat #supercollider-development to talk/coordinate about developer topics