On Tue, 12 Apr 2005, [iso-8859-1] G=E1bor Melis wrote:
> I would say that with the threaded version the individual finalizers=20
> need only be thread safe (as opposed to reentrant).
Good point. Provided that the second call occurs in another thread this is=
indeed the case; and indeed I think this was something David tried to=20
bring home to me a couple of days back on #lisp. I seem to be a bit slow=20
on the pickup...
One question for those of you running threaded systems: could someone=20
outline pros and cons for a separate finalizer thread?
-- Nikodemus Schemer: "Buddha is small, clean, and serious."
Lispnik: "Buddha is big, has hairy armpits, and laughs."