changeset in /hg/icedtea6: Backport newly Freed tests.

[Joseph D. Darcy]

Andrew John Hughes wrote:
> On Fri, 2009-02-13 at 12:49 -0800, Joseph D. Darcy wrote:
>   
>> Andrew John Hughes wrote:
>>     
>>> 2009/2/13 Joseph D. Darcy <Joe.Darcy at sun.com>:
>>>   
>>>       
>>>> Hello.
>>>>
>>>> The annotations test patch is unnecessary.
>>>>
>>>> The tests in question that in JDK 7 live in
>>>>   test/java/lang/annotation (singular)
>>>> have since build 13 lived in OpenJDK 6 at
>>>>   test/java/lang/annotations (plural)
>>>>
>>>> See
>>>> http://hg.openjdk.java.net/jdk6/jdk6/jdk/file/f6b99ef3deee/test/java/lang/annotations/
>>>>
>>>> The original closed tests lived in an "annotations" directory, but by the
>>>> naming conventions for tests they should have been in "annotation" so I put
>>>> the JDK 7 tests in the latter directory when I opened them up for that
>>>> release.
>>>>
>>>>     
>>>>         
>>> Ok, this wasn't mentioned in the commit or anywhere else.
>>>       
>> Why would it be?
>>
>>     
>
> Why wouldn't it be?  

There is an arbitrary amount of additional information that could be 
included in a commit message.

> Sun commits seem to lack any comments other than
> bare bug IDs.  

Yes, all commits require a bug id and the bug database is expected to 
hold most of the information about the bug.

> If the motivation for patches was included, along with
> known issues for this, it would make working with them a lot easier.
> IcedTea commits at least have a ChangeLog, though I would like commits
> there to be more verbose as well.
>
> There is a review process for these commits but it seems it is still
> completely internal to Sun.
>
>   
>>>   Can you fix
>>> the naming in OpenJDK6 to match?
>>>   
>>>       
>> I do not plan to move the existing OpenJDK 6 tests.
>>
>>     
>
> Is there any reason to keep this arbitrary difference, given it will
> create headaches merging between the two?
>   

I do not expect these tests to evolve very much.

>>>> Especially since there is a public OpenJDK 6 Mercurial forest now, in the
>>>> future when there is interest in having a Sun patch ported to IcedTea 6, I
>>>> suggest contacting us at Sun about getting the change in the upstream
>>>> OpenJDK 6 master directly.
>>>>
>>>>     
>>>>         
>>> I expect these tests will be in the next build drop.  The point of
>>> backporting them was so we could start running them in the meantime.
>>>
>>>   
>>>       
>> If the goal is having more tests (or other changes) in the upstream 
>> OpenJDK 6 repository there are two general paths:
>>
>> 1) Apply patch to IcedTea6
>> 2) Changes later get added to OpenJDK 6 directly
>> 3) Remove newly unnecessary patch from IcedTea6
>>
>> or
>>
>> 1) Change OpenJDK 6 directly
>>
>>
>> If it happens promptly, the second path strikes me as more efficient all 
>> around :-)
>>
>>     
>
> I agree, except it's actually:
>
> 1) Change OpenJDK6 directly
> 2) Change appears in a build drop
> 3) IcedTea6 picks up the new build drop
>
> i.e. we base IcedTea on the latest build drop, not hg. 

That was certainly necessary before OpenJDK 6 had live hg.

>  Not only is hg
> slower, but build daemons don't generally have web access (this is also
> an issue for testing) and we need a clear snapshot to work against.
>   

Each revision of the repository is identifiable of course.

> 1 is also still held back by the current lack of external committers.
>   

That should change over time.

-Joe