On 29/01/2013 23:36, David Buck wrote:
Hi!
This is a review request to add only the test case for the following
OracleJDK issue :
[ 7042126 : (alt-rt) HashMap.clone implementation should be re-examined ]
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7042126
* please note: I just marked the bug as "public" this morning, so
there will be a short delay before the above link is available.
The issue (root cause) is not in OpenJDK (i.e. the problem was
OracleJDK specific), but the test case is valid for any Java SE
implementation so it should go into OpenJDK so we can prevent a
similar issue from ever happening in both releases moving forward. The
test case simply helps ensure that the contents of a HashMap are not
leaked when we clone the HashMap.
webrev:
[ Code Review for jdk ]
http://cr.openjdk.java.net/~dbuck/7042126/webrev.00/
How robust is this test? I'm concerned that it might fail intermittently
if the System.gc doesn't immediately GC the now un-references entries in hm.
Minor nit, should be WeakReference<Object> to avoid the raw type.
-Alan.
