Subj : Re: SpiderMonkey crashes on invalid JSAtom*
To   : netscape.public.mozilla.jseng
From : "itaj sherman" <myngs@hotmail.com>
Date : Mon Feb 24 2003 05:55 pm

in property hooks (get, set, resolve) we don't invoke JS_GC directly but we
use JS_DefineProperty, JS_NewObject and other API that may invoke the GC
internally.
as i said, by the GCCallback i set the GC to be active only in the special
GC thread, but there's this section in ClaimScope() that suspends the
request and may let the GC execute in the GC thread while still in the
property hook.
do you think it could affect that?

"Steven C. Cole" <scole@planetweb.com> wrote in message
news:b30k7g$e1c1@ripley.netscape.com...
> itaj sherman wrote:
> > hi,
> >
> > i'm experiencing crashes in my application that have to with the atom
table
> > that SpiderMonkey keeps on the JSRuntime.
>
> Could you be running a GC from within a resolve hook (or any other hook
> for that matter)?   The atom table isn't rooted during resolve, and a
> JS_GC there will blow it away, leading to the type of problem you're
> experiencing.  (It happened to us...)
>
> --scole
>

.