Ah, I think I found out why it's not a problem. The spurious
IdentityWeakReference instances are indeed added to the queue. But once
they are ready for reaping, they are set to null already. So no live
objects can ever get lost.
Still, it would be a tiny efficiency win not to add references to the
Hi, I'm from the "Fix Java" department.
This is an advanced problem, please bear with me.
The issue concerns WeakIdentityHashMap (
https://github.com/apache/avro/blob/master/lang/java/avro/src/main/java/org/apache/avro/util/WeakIdentityHashMap.java).
I believe there may be a bug.
Suppose the rem
[
https://issues.apache.org/jira/browse/AVRO-1749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16285265#comment-16285265
]
Sebastien Dubois commented on AVRO-1749:
Any news over this feature request & PR?
I