Subj : Re: Narrowed the problem down to... To : netscape.public.mozilla.jseng From : "Sterling Bates" Date : Tue Nov 04 2003 04:37 pm "Brendan Eich" wrote in message news:3FA8354D.60601@meer.net... > Sterling Bates wrote: > > >If I comment this line out, msvcr70.dll is happy. As soon as it's > >reintroduced, a breakpoint is fired when destroying the context. > > > > What's the stack? That's the problem...Delphi won't show a stack trace into the DLL. I can manually step through the machine code, but I have a very hard time keeping track of where I am. All I can say is that I've traced as far as a call to free the DHash table(?). My head starts to hurt after that :) I know I'm not able to provide much for troubleshooting -- it's frustrating on my end too. FWIW, I've done scads of searches on Google, but none of my searches produce much. Of course, I don't think it's Delphi per se, since other objects have worked in the past. The only difference is this one is instantiated within an object. > What exactly is that context? You say it's a "running context from > another object" -- what does that mean? Is there more than one thread > involved here? You can't use a context on two threads at the same time, > of course. No, just one thread. One object inits a JSRuntime and a JSContext, and I use a reference to the JSContext in the other object. > It's almost never the case that a caller "hands off" ownership of a > string to a callee. This case is typical: the caller owns the string, > the callee will have to make a copy if it needs one (but JS doesn't use > char, it uses jschar for Unicode, so any "copy" made by > JS_DefineProperty or a subroutine of it won't be a duplicate char > string, anyway). That's kinda what I thought. Oh well. Thanks, Sterling .