Crash on attempt to import non-GM instrument on a case of bank absence

[Wohlstand]

When you have count of bank less than importing instrument, the crash will happen as the reason of instruments store arrays mismatch. To fix this, it's need to check the instrument number, and then automatically expand the banks set to let instruments be correctly imported.

P.S. It's absolutely same crash on OPL3 Bank Editor as the same importer code was used in both cases.

[jpcima]

I think I understand about the mismatching banks.
What's the deal about the instrument vector and the pointer as distinct members? can't an instance of bank keep only the vector?

[Wohlstand]

What's the deal about the instrument vector and the pointer as distinct members? can't an instance of bank keep only the vector?

It's easy:

Every time you are resizing the Ins_Melodic_box vector, you must re-assign the Ins_Melodic with Ins_Melodic_box.data(); to renew the pointer that was became dead. Same with the case of drums.

However, I think, I'll equip the "FmBank" with generic functions to resize instruments storage without of direct deal with vector and pointer.

[jpcima]

This was not sufficient. When you import by "assign" mode, you have to match the banks of destination and source.
I think it means to create a bank if it does not exists in the destination, and dstination instrument can have a different index in the store.

[Wohlstand]

and dstination instrument can have a different index in the store.

Therefore don't do 1:1 assignment, but use MSB/LSB pair to identify the target bank (but, prevent ability to set duplicated bank key, the pair of MSB/LSB must be unique).

[Wohlstand]

I think, it's fine now. The fix that was taken long time ago, is fine 😉