El mar, 21-07-2009 a las 01:13 -0400, Schwab,Wilhelm K escribió:
> Hello all,
>
> I have a weird-looking bug that I am willing to assume is my doing and in my
> code. That said, some debugging lead me to a situation I really do not
> enjoy: objects that I think should be gc'd are happily living in my image. I
> _think_ I have gagged workspace variables (they truly are more trouble than
> they are worth). Dolphin sometimes has "zombie" views that can lurk in a
> broken/invisible state but still hold onto other objects. Is there anything
> like that here? I know about the various #become: tricks, and do not like to
> resort to them; I would like to understand what is holding onto the would-be
> garbage.
>
> Can the Squeak/Pharo gc collect objects with cyclic references? Is there
> anything else comes to mind when hunting for objects that just won't go away?
>
> Bill
Use the pointer finder, either from a debug window where you can see a
variable that you think should not exist, search the context menu for
"Chase pointers" or "explore pointers" or directly by using
PointerFinder. From the class comment:
I can search for reasons why a certain object isn't garbage collected.
I'm a quick port of a VisualWorks program written by Hans-Martin Mosner.
Call me as shown below. I'll search for a path from a global variable
to the given object, presenting it in a small morphic UI.
Examples:
PointerFinder on: self currentHand
PointerFinder on: StandardSystemView someInstance
Now, let's see why this image contains more HandMorphs as expected...
HandMorph allInstancesDo: [:e | PointerFinder on: e]
Miguel Cobá
>
>
> _______________________________________________
> Pharo-project mailing list
> [email protected]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________
Pharo-project mailing list
[email protected]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
