Code review request: 7077672 jdk8_tl nightly fail in step-2 build on 8/10/11

Jonathan Gibbons jonathan.gibbons at oracle.com
Thu Aug 11 23:29:22 UTC 2011


On 08/11/2011 02:56 PM, Rémi Forax wrote:
> On 08/11/2011 10:55 PM, Jonathan Gibbons wrote:
>> Rémi, you wrote:
>>
>>> On 08/11/2011 09:07 PM, Alexandre Boulgakov wrote:
>>> >/  Please review the attached patch to fix a build break.
>>> />/
>>> />/  The fix changes new Class<>[] to new Class<?>[] in two places.
>>> />/
>>> />/  Thanks,
>>> />/  Sasha
>>> /
>>> Hi Sasha,
>>> you can rewrite:
>>>
>>>     connectMethod = corbaStubClass.getMethod("connect",
>>>                   new Class<>[] {org.omg.CORBA.ORB.class});
>>>
>>> to
>>>
>>>     connectMethod = corbaStubClass.getMethod("connect",
>>>                   org.omg.CORBA.ORB.class);
>>>
>>> because getMethod is a varargs.
>>>
>>> Anyways, the changes are ok for me without that.
>>>
>>> Rémi
>>
>> Thanks for the suggestion.  In this case we wanted a minimal fix to 
>> go back as safely and as quickly as possible.  But your suggestion 
>> leads to another, broader suggestion for a cleanup of places where 
>> the array declaration is no longer required.  Another one for the list!
>>
>> -- Jon
>
> Hi Jon,
> first I'm really happy of these clean up,
> I used to use the JDK source code as an example of well written code 
> for my students,
> so any improvements is great.
>
> Also I agree with you, the proposed change above is not as important as
> the changes Sasha is doing because it just improves the readability
> not the type safety.
> So it can be done later.
>
> cheers,
> Rémi
>

Rémi,

"I used to use ...." ?    :-(

If you see any more opportunities for cleanup, I hope you'll let us 
know.  Patches (or even changesets) would be even better :-)   You could 
even have fun with the compiler and tree API to find examples of unclean 
code that are candidates for cleanup.  After all, why should we have all 
the fun?

-- Jon



More information about the core-libs-dev mailing list