Re: releasing a thread object (Gworkspace)

[Riccardo Mottola]

Hi,

Riccardo Mottola wrote:
> I suppose this means the remote object is already dead or somehow 
> dealloc'ed, even if i didn't see a dealloc. 
actually, by chance, I found out that just by changing the protocol 
definition of "endOperation" of the master obejct (a method that the 
task calls at the end to do some cleanup) from oneway to synchronous 
fixed the problem and I see a dealloc,

Done this, I am able to perform some further cleanup. I see that in on 
thread exist I was not always removing notification observer (e.g. 
threadWillExit should match detachOperationThread)

Thanks for the support, if I have other questions about this cleanup. 
Now I need some stress-testing.

Riccardo

Index: FileOpInfo.h
===================================================================
--- FileOpInfo.h        (revision 38124)
+++ FileOpInfo.h        (working copy)
@@ -45,7 +45,7 @@

 - (void)removeProcessedFiles;

-- (oneway void)endOperation;
+- (void)endOperation;