Wed, 20 Feb 2008 16:38:49 +0100ProxyLookup.toString was broken release61-beta_base
Jaroslav Tulach <jtulach@netbeans.org> [Wed, 20 Feb 2008 16:38:49 +0100] rev 346
ProxyLookup.toString was broken

Tue, 19 Feb 2008 07:59:28 +0100Merging with trunk, removing the original fix for re-entrancy and its tests replacing it with complete new one based on immutable data structure
Jaroslav Tulach <jtulach@netbeans.org> [Tue, 19 Feb 2008 07:59:28 +0100] rev 345
Merging with trunk, removing the original fix for re-entrancy and its tests replacing it with complete new one based on immutable data structure

Thu, 14 Feb 2008 16:42:24 +0100Initializing the class sooner, not using atomic boolean
Jaroslav Tulach <jtulach@netbeans.org> [Thu, 14 Feb 2008 16:42:24 +0100] rev 344
Initializing the class sooner, not using atomic boolean

Thu, 14 Feb 2008 12:01:58 +0100Replacing annonymous classes with named ones. This way we hopefully find out what is the cause for the test to not be able to load $11 on the testing machine...
Jaroslav Tulach <jtulach@netbeans.org> [Thu, 14 Feb 2008 12:01:58 +0100] rev 343
Replacing annonymous classes with named ones. This way we hopefully find out what is the cause for the test to not be able to load $11 on the testing machine...

Wed, 06 Feb 2008 19:33:00 +0100So finally I know who exits from the critical section into foreing code. It is res.lookup(template) in lookupChange. This fix ensures that res.lookup(template) is never called with lock being held on the ProxyLookup. The fix may not be perfect, but at least it should guarantee internal consistency of the ProxyLookup's own data structures.
Jaroslav Tulach <jtulach@netbeans.org> [Wed, 06 Feb 2008 19:33:00 +0100] rev 342
So finally I know who exits from the critical section into foreing code. It is res.lookup(template) in lookupChange. This fix ensures that res.lookup(template) is never called with lock being held on the ProxyLookup. The fix may not be perfect, but at least it should guarantee internal consistency of the ProxyLookup's own data structures.

Tue, 19 Feb 2008 07:41:29 +0100eliminating the weak_proxy_lookup_but_does_not_GC_template_and_its_class head
Jaroslav Tulach <jtulach@netbeans.org> [Tue, 19 Feb 2008 07:41:29 +0100] rev 341
eliminating the weak_proxy_lookup_but_does_not_GC_template_and_its_class head

Tue, 05 Feb 2008 19:43:00 +0100An attempt to make the result hold the data weak, but it does not work because of GC of template's class weak_proxy_lookup_but_does_not_GC_template_and_its_class
Jaroslav Tulach <jtulach@netbeans.org> [Tue, 05 Feb 2008 19:43:00 +0100] rev 340
An attempt to make the result hold the data weak, but it does not work because of GC of template's class

Tue, 19 Feb 2008 07:23:16 +0100Getting rid of Xlint warning
Jaroslav Tulach <jtulach@netbeans.org> [Tue, 19 Feb 2008 07:23:16 +0100] rev 339
Getting rid of Xlint warning

Tue, 19 Feb 2008 07:16:25 +0100Giving up, returning back the finalizer to make testGarbageCollect pass
Jaroslav Tulach <jtulach@netbeans.org> [Tue, 19 Feb 2008 07:16:25 +0100] rev 338
Giving up, returning back the finalizer to make testGarbageCollect pass

Mon, 18 Feb 2008 20:45:21 +0100Reorganized, support for cleanup of templates, but the classloader is still not GCed
Jaroslav Tulach <jtulach@netbeans.org> [Mon, 18 Feb 2008 20:45:21 +0100] rev 337
Reorganized, support for cleanup of templates, but the classloader is still not GCed