Christian Müller ha scritto:
> 
> 
> Andrea, my hero, you fixed http://jira.codehaus.org/browse/GEOT-2546 
> successfully. What have you done ?

The test did not force the map to have a number of hard references equal
to the sample size, so basically it relied on the random distribution
of add/remove to never have 20 more adds than removes at any time.
The Sun boolean random distribution must be a lot more uniform than
the IBM one, even in the short term, so it never failed there.

> I seems there is a twin of the problem
> Failed tests:
> testWeakReferences(org.geotools.util.CanonicalSetTest)
> and sometimes
> testStrongReferences(org.geotools.util.CanonicalSetTest)
> Last time the problem occurred  building geotools 2.6.x with ibm sdk 5 
> 64 Bit, and it is not reproducible within eclipse.
> What was the problem with GEOT-2546 ?. Perhaps we have a similar 
> situation here.

Perhaps, but the canonical set is made only of weak references.
In theory having the other hash hold a strong reference onto the
same objects should prevent the weak one to loose references.

Unfortunately it's not reproducible in Eclipse, as you said, so
I have no idea of why it's failing...

Cheers
Andrea

-- 
Andrea Aime
OpenGeo - http://opengeo.org
Expert service straight from the developers.

------------------------------------------------------------------------------
This SF.Net email is sponsored by the Verizon Developer Community
Take advantage of Verizon's best-in-class app development support
A streamlined, 14 day to market process makes app distribution fast and easy
Join now and get one step closer to millions of Verizon customers
http://p.sf.net/sfu/verizon-dev2dev 
_______________________________________________
Geotools-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-devel

Reply via email to