Re: Segfault at quit with global and DLD type

[Jason Riedy]

And Michael Goffioul writes:
> I don't think it's really a race condition, but more the order in
> which objects are destroyed.

Sorry, I used "race condition" in a general sense to mean that order.

> More precisely, I think (John, could you confirm?) that DLD functions
> are cleared before global variables.

That would be it, thanks. I'll rewrite a few dozen routines to
work around the problem... Sounds awful, but it needs done
anyways to refactor the common bits.

I'm not sure there is a sensible way around the problem short
of never clearing oct files while related variables are
active.  Hm.  Hacking up the type table to count references to
loaded types might work.

Jason