Hi! There was a disussion in the mailing list under the "[VM]How to trigue GC while free native memory is low." and "[VM]take action for native memory control (was: Re: [VM]How to trigue GC while free native memory is low.)" topics discussing the problem of freeing native memory in finalizers - see http://comments.gmane.org/gmane.comp.java.harmony.devel/23680 and http://comments.gmane.org/gmane.comp.java.harmony.devel/24014 for example.
Leo Li provided the patch for http://issues.apache.org/jira/browse/HARMONY-3148 as it was discussed in the thread. (Thanks, Leo!) I've reviewed it and it looks great to me. There is also http://issues.apache.org/jira/browse/HARMONY-2530 JIRA about memory leak in java.util.zip.Deflater (and i've check that similar problem exists in java.util.zip.Inflater although there are no JIRA regarding this) which could be fixed using the facility suggested by Leo's patch after the integration for 3148. I think it makes sense to integrate the patch for 3148. Any objections? Could also someone from committers take care of this? Thanks, Mikhail
