Code review request: 7085024 internal error; cannot instantiate Foo

Rémi Forax forax at univ-mlv.fr
Thu Oct 6 06:52:47 PDT 2011


On 10/06/2011 03:25 PM, Jonathan Gibbons wrote:
> On 10/06/2011 05:01 AM, Rémi Forax wrote:
>> On 10/06/2011 12:50 PM, Maurizio Cimadamore wrote:
>>> On 06/10/11 11:01, Rémi Forax wrote:
>>>> On 10/06/2011 11:47 AM, Maurizio Cimadamore wrote:
>>>>> On 29/09/11 01:07, Dmeetry Degrave wrote:
>>>>>> http://bugs.sun.com/view_bug.do?bug_id=7085024
>>>>>>
>>>>>> fix is identical to jdk8, this is a back port to jdk7u.
>>>>>>
>>>>>> webrev : http://cr.openjdk.java.net/~dmeetry/7085024/webrev.00/
>>>>>> jdk8: http://hg.openjdk.java.net/jdk8/tl/langtools/rev/b86277584776
>>>>>>
>>>>>> thanks,
>>>>>> dmeetry
>>>>> Approved
>>>>>
>>>>> Maurizio
>>>>
>>>> if botFilter is declared static final,
>>>> the patch is Ok for me.
>>> final yes
>>> static no :-)
>>>
>>> it is supposed to be used if you have an instance of Types around 
>>> [and the context instantiation protocol will already ensure there's 
>>> only _one_ instance of Types]
>>>
>>> Maurizio
>>
>> That's true for javac, not for any tools that will use javac classes but
>> not its kind of dependency manager.
>
> See the comment at the head of most javac source files.

I know, I know.

>
>>
>> Anyway, this code will be transformed into a lambda at the start of 
>> JDK9 timeframe :)
>
> Maybe, but don't hold your breath. javac is deliberately conservative 
> in the language features it uses.

It depends on the seasons. At some point in the past, javac was the only 
program
that was using generics.

Rémi




More information about the compiler-dev mailing list