Need reviewers for changes in java/net regression tests
joe.darcy at oracle.com
Mon Aug 2 19:25:03 PDT 2010
Dr Andrew John Hughes wrote:
> On 18:49 Thu 29 Jul , Joe Darcy wrote:
>> Chris Hegarty wrote:
>>> On 07/26/10 14:00, Dr Andrew John Hughes wrote:
>>>> I'm afraid I don't see the issue with bug IDs here. Each OpenJDK7
>>>> changeset already has an associated bug ID. It should thus simply be
>>>> a matter of backporting the relevant changesets and dropping any
>>>> changes from them that should remain 7 only.
>>> Some of the test changes have been made when fixing a bug in the JDK7
>>> source... Oh, I noticed that the test in this area is not sufficient
>>> or has issues. Some tests have been fixed as part of a general cleanup
>>> to remove them from the ProblemList. We don't want to backport all the
>>> implementation source changes, only test changes that improve the
>>> quality of the tests. So there is no one to one mapping here.
>> If there are fixes for correctness issues allowable under the Java SE 6
>> specification, I'd be open to the fixes being backported to OpenJDK 6.
> I still think that, even if only a subset of a changeset is backported, the
> same bug ID should be used. Creating huge changesets which don't have a clear
> trace back to 7 makes things very hard to track.
I agree with that to a point. If there is a behavioral fix and test
changes including test refactorings, just the test refactoring should
not be ported under the same bug id since that would imply the behavior
change was included as well. (If there is a behavioral change and the
revised test passes with or without the behavioral change, the tests for
that fix are inadequate IMO.)
More information about the jdk6-dev