Comment 23 for bug 2036817

(In reply to Michael Weghorn from comment #16)
> It's still possible/likely that other crashes on exit for different
> scenarios with a similar backtrace will still occur with that fix in place
> since some change now seems to trigger those when a11y objects haven't
> properly been disposed before, i.e. pre-existing issues with the object
> lifecycle now apparently result in a crash, while they didn't before.
>
> It might be possible to identify/bisect (actual bisect, not using pre-built
> binaries from bibisect repo) what change causes this, but for now I'd rather
> focus on fixing the root causes if more of these issues pop up.
> (IIUC, there shouldn't be any problem of data loss, since the crash is on
> exit, i.e. after docs have been saved already).

@Noel: FYI. Any opinion/further thoughts? (Might be related to your previous changes for fixing a11y-related crashes on shutdown.)