Question about MethodContext, tempNames and GC

Hi. These days, I needed to implement a loop detector for one of my projects, and I started to play a lit with thisContext. I've used it before but not as deep as now. It is AWESOME!!!!

Now...I noticed (and I think this was the most awesome thing) that I can not only access to senders and their receivers and methods, but also to the arguments!!!
And then I thought... mmmmmm  how can I access the arguments of a sender of a sender of .... because I thought they should be garbage collected, but if I can access. Ok, there is when I discover that this information is stored in a DebuggerMethodMap and  I guess this guy uses a kind of WeakArray to store them.

Now, for my loop detector, I use  thisContext to access senders AND I need to access the arguments of a method of a sender. For that I use the method ContextPart >> tempsAndValues
The question is....isn't this method quite "fragile"? I mean, doesn't this work ONLY if a GC was not run between thisContext and that context ?  I am right ? I need to know this to see if I can trust on this or not.
And if I am right, maybe we can modify and add this info in a comment?

Thanks

Mariano

> And then I thought... mmmmmm  how can I access the arguments of a sender of
> a sender of .... because I thought they should be garbage collected, but if
> I can access. Ok, there is when I discover that this information is stored
> in a DebuggerMethodMap and  I guess this guy uses a kind of WeakArray to
> store them.

No, arguments need to kept with the context object until the context
object itself is not used anymore (e.g. not referenced from anywhere
and execution completed with a return or unwind).

> Now, for my loop detector, I use  thisContext to access senders AND I need
> to access the arguments of a method of a sender.

Go to the Process Browser and click on 'start CPUWatcher'. It kind of does that.

Lukas

--
Lukas Renggli
www.lukas-renggli.ch