Subj : Re: JS_HandleTrap unhandled exception
To   : netscape.public.mozilla.jseng
From : Brendan Eich
Date : Fri Apr 01 2005 09:11 am

Mike Moening wrote:
> Ok...  Even IF I managed to lock the trap and watchpoint lists (which
> are...oh rats...at the runtime level) I still don't think it will work
> unless the JSHandleTrap function becomes more forgiveing and returns
> JSTRAP_CONTINUE when FindTrap returns NULL because the trap was cleared.
> That is unless the JSOP_TRAP is also taken out when JSClearTrap is called
> (didn't have time to check this but I doubt it)


You're right, and I misremembered the design.  Again jband is the man -- 
he wrote the first JS debugger on top of jsdbgapi.[ch], and the design 
limit was always a single-threaded debugger.


> The debug controller we are using is meant to be fully multi-threaded.  i.e
> as many debuggers as you want as long as they are each attached to a separt
> context.
> Must embedders go the MT way.  We really want to support that.


Right, but debugging is usual a serial task with a human in the loop, 
where you want to stop all threads.  That's why we reckoned, so long ago 
(was it 1997 or even late 1996?) that the single-threaded debugger 
design limit was acceptable.


> I think I could pretty easily make this baby thread safe if you don't mind
> me making the changes...


You could try, but if we end up trading patches and code reviews, and 
rewriting by review, I may just grab it.

/be

.