Re: lv2 support for Debian 12 & zooming crash
Ted Felix <ted-Ll1PANez7xxWk0Htik3J/[email protected]> Wed, 21 Jan 2026 14:12:53 -0500
| Newsgroups | gmane.comp.audio.rosegarden.user |
|---|---|
| Message-ID | <[email protected]> |
> If this is a bug and very rare, will it be named after me? Ha! No. It will be named after whatever part of Qt is messing this up. Give the latest in git a try. I've rewritten all the matrix zoom code and it might be more stable for you now. Ted. On 1/21/26 6:52 AM, Babarosa wrote: > Do you have a procedure that starts from an empty composition? > > Not really, I didn't manage to crash Rosegarden using an empty composition. > > Please do not laugh, but to trigger this crash I use the zoom wheels very rapidly like a lunatic (!) from maximum to minimum. Sometimes Rosegarden crashes after 15 movements, sometimes it takes 30 seconds. > > If this is a bug and very rare, will it be named after me? > > Thanks and Greetings, Michael > > >> Gesendet: Montag, 19. Januar 2026 um 03:25 >> Von: "Ted Felix" <ted-Ll1PANez7xxWk0Htik3J/[email protected]> >> An: Babarosa <[email protected]>, rosegarden-user <rosegarden-user-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org> >> Betreff: Re: [Rosegarden-user] lv2 support for Debian 12 & zooming crash >> >> Fascinating. This crash only touches us once at Panned.cpp:66. And >> that's just a message forward. Unlikely that it is an rg problem. More >> likely a Qt5 bug. We might want to try watching for a null event >> pointer there, but I have a feeling that never happens as it would cause >> an earlier crash (shorter call stack). >> >> Do you have a procedure that starts from an empty composition? Don't >> spend any time on it if not. It would just be nice to see if it is >> reproducible with Qt6 and Philip could test that very easily. >> >> I'll spend a little time beating it up with some huge compositions I >> have lying around. >> >> Ted. >> >> On 1/18/26 5:24 PM, Babarosa wrote: >> > Please find attached the info. I can reproduce this crash. >> > >> > Thank you for helping, Michael >> > >> > Am 18.01.26 um 16:43 schrieb Ted Felix: >> >> Ok. Usually for this kind of thing, running a debug build using ASAN >> >> will give us a lot more information when it does crash. Or you can >> >> run a debug build with gdb, which is a little more cumbersome, but >> >> should get some good crash information for us to work with. >> >> >> >> Let me know if you need help setting either of those up. The cmake >> >> line for ASAN is: >> >> >> >> cmake .. -DCMAKE_BUILD_TYPE=Debug -DCMAKE_CXX_FLAGS="-Werror - >> >> fsanitize=address -fno-omit-frame-pointer" >> >> >> >> It's in the README.md. >> >> >> >> Ted.