Message8823

Author zyasoft
Recipients indra, rchyla, zyasoft
Date 2014-06-25.06:22:14
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1403677335.17.0.690204552765.issue2127@psf.upfronthosting.co.za>
In-reply-to
Content
Applied Indra's patch in https://bitbucket.org/jimbaker/jython-resource-leaks, with the intent to merge this into trunk sometime soon, maybe this week

Note that there are a number of failing tests. I don't think we need to support recursion_limit, that's clearly a CPython-implementation detail. It's interesting to see that the change in ThreadStateMapping to use a weakkey/weakval map instead of a ThreadLocal is enough to cause a failure in test_weakset. However, I really doubt that _IterationGuard is as threadsafe as it claims, at least as it interacts with the remove method, given that 

                if self._iterating:
                    self._pending_removals.append(item)
                else:
                    self.data.discard(item)

is clearly not guaranteed to be atomic. The solution is to revisit this implementation with the use of Guava MapMaker.
History
Date User Action Args
2014-06-25 06:22:15zyasoftsetmessageid: <1403677335.17.0.690204552765.issue2127@psf.upfronthosting.co.za>
2014-06-25 06:22:15zyasoftsetrecipients: + zyasoft, indra, rchyla
2014-06-25 06:22:15zyasoftlinkissue2127 messages
2014-06-25 06:22:14zyasoftcreate