DEBUG ASSERT: "mappingFile.exists()" in LegacyControllerMapping::loadSettings()

[mxmilkiib]

Bug Description

Removed unfinished script files, tried various ChatGPT generated files which didn't work, tried changing back to parts of older files in steps, couldn't even see in the preset menu, moved .mixxx/controller dir and made a new one, tried to use the Learning Wizard, crash.

https://gist.github.com/mxmilkiib/d7fa17e801ec6c986ba256c8f1b48164 down to afair 2136

DEBUG ASSERT: "mappingFile.exists()" in function void LegacyControllerMapping::loadSettings(UserSettingsPointer, const QString&) const at ./src/controllers/legacycontrollermapping.cpp:14

Version

2.6

OS

Linux

[acolombier]

This is a debug assert - you may continue the execution upon assert (press c in GDB when getting a SIGINT). You may disable debug assert if you aren't developing the Mixxx core and don't want to be bother. I can confirm that the loadSetting is not expected to be called in your scenario so this could use some improvement although looking at the code I don't think it should cause a crash.

[mxmilkiib]

Just rebuilt Mixxx with a clear cmake and it also coredumps.

[mxmilkiib]

This happens for any controller when clicking the Learning Wizard.

[ronso0]

IIUC this stems from removing mapping files while they are still considered 'loaded', ie. the config still holds the mapping file path? This can (but shouldn't) happen, so shall we turn the DEBUG_ASSERT into VERIFY_OR_DEBUG_ASSERT, or even just if (!mappingFile.exists())?

@acolombier Are you already working on this / plan to do so? If no, I'd take a stab.

[acolombier]

Sounds like the fix! Haven't looked at it yet, please go ahead @ronso0

[ronso0]

Yup, see #13738

Btw I can confirm we hit the assert when starting the Learning Wizard.