Okay, I'm looking at this... the assertion happens because w_current->inside_action gets cleared by something to do with pressing Ctrl+X. I'm not even sure why that assertion is there, TBH, because the function doesn't seem to do anything that requires inside_action to be non-zero... but we probably should try and fix this anyway.
I really wish we had state-dependent keybindings -- I suspect there are a lot of nasty surprises like this one tucked away because no-one checks (no-one *can* possibly check) every possible combination of different actions taken in different editing states...
Okay, I'm looking at this... the assertion happens because w_current- >inside_ action gets cleared by something to do with pressing Ctrl+X. I'm not even sure why that assertion is there, TBH, because the function doesn't seem to do anything that requires inside_action to be non-zero... but we probably should try and fix this anyway.
I really wish we had state-dependent keybindings -- I suspect there are a lot of nasty surprises like this one tucked away because no-one checks (no-one *can* possibly check) every possible combination of different actions taken in different editing states...