switching puzzles from 2360 causes a crash

Started by LociOiling

LociOiling Lv 1

I'm getting mixed results reproducing the crash. A second client (already running) also crashed. A third client, that was not running, updated to the new main, then didn't crash when switching from 2360 to 2358. The two clients that crashed can now switch from 2360 to 2358 without a problem.

Yet another client (the fourth), which was already running, also experienced the crash:

log.txt

After restarting, that fourth client is also able to switch from 2360 to 2358 without a crash.

LociOiling Lv 1

Major edit, the ligand queue has been exonerated. It's actually the small molecule properties that cause the crash.

To recreate, select the ligand, the hit "v" for the view options. Select one of the options on either of the two panels. Now Foldit crashes if you attempt to switch to puzzle 2358.

One of the visual options, such as "pi pi interactions" is enough to produce the crash if it's still selected when you switch. If you uncheck the option, no crash occurs.

The numeric values, such as "Mcl weight" seem to follow the same pattern. No crash if none are checked, but a crash if any are checked.

At least in some cases, opening the ligand queue seems to be involved in the crash. (Edit: the crash rate is still not 100% however.)

~~If the client hasn't opened the ligand queue since the restart, switching from 2360 to 2358 does not crash. ~~

If the ligand queue has been opened, Foldit crashes switching from 2360 to 2358. (Edit: somewhat reliably anyway.)

Since the ligand queue is "modal", you have to close it before you can open the puzzle menu (also modal). Closing the ligand queue once it's been opened doesn't prevent the crash.~

Yet another log file attached.

log.txt

LociOiling Lv 1

Yet another follow-on, this is not a new problem. When I load expired puzzle 2216, switching to puzzle 2358 causes a crash if any of the small molecule properties are selected.

rmoretti Staff Lv 1

Yeah, it looks like the Ligand Properties Panel isn't properly getting deactivated on switching puzzles (or more precisely, it isn't getting deactivated in the correct order), so it tries to grab the current ligand from the new puzzle, but that crashes (because it no longer exists).