Question re: JEP 492 specification

[Archie Cobbs]

OK thanks to both of you for clarifying this. I agree it's good to keep the
rule simple.

I realize now I was kind of thrown off by the lack of connection with
flexible constructors and the fact that this compiles in (e.g.) JDK 17, but
that was because I didn't read the little note carefully enough...

*This missing condition is a bugfix for extensions introduced in JEP 395.*

Thanks,
-Archie

On Tue, Dec 10, 2024 at 7:21 PM Maurizio Cimadamore <
maurizio.cimadamore at oracle.com> wrote:

> What Chen said.
>
> In principle, in this case, creation could be allowed because the local
> doesn’t capture. But when discussing the rules, we reached the conclusion
> that we didn’t want to specify what “capture” meant - or classify local
> classes in terms of what they happen to be capturing. So, if that is what
> you were asking, I’d say the behavior you are observing is deliberate.
>
> Dan correctly pointed out, during an off-line discussion, that what it
> would be useful, in some cases, would be the ability to declare a local
> *static* class. In which case you are stating that capture of variables
> in the enclosing context by the class is just not a possibility. In that
> case, the rules can be relaxed, but on a much more principed basis.
>
> Maurizio
>
> On 10/12/2024 22:47, Chen Liang wrote:
>
> Hi Archie,
> I think this rule makes sense - Local1 can capture local variables in the
> outer class initializer, in which case the error will be legitimate. The
> extra otherwise does not fix this issue. Adding a special case to allow
> no-capture local classes to be used in more nested static contexts seems
> overkill and error-prone.
>
> Chen
>
> On Tue, Dec 10, 2024, 3:57 PM Archie Cobbs <archie.cobbs at gmail.com> wrote:
>
>> The proposed JLS changes for JEP 492 are saying that this example should
>> no longer compile*:
>>
>>     static {
>>         class Local1 {
>>             class Local2 {
>>                 public static void m() {
>>                     new Local1();  // error
>>                 }
>>             }
>>         }
>>     }
>>
>> but that seems wrong - that exampple has been allowed since JDK-8254321,
>> and as it has nothing to do with flexible constructors, it should continue
>> to be allowed.
>>
>> The JEP 492 spec says (new additions in bold):
>>
>>   If *C* is an inner local class, then:
>>
>>    -
>>
>>    If *C* occurs in a static context, then *i* has no immediately
>>    enclosing instance. *Let S be the nearest static method declaration,
>>    static field declaration, or static initializer that encloses the
>>    declaration of C. If the nearest static method declaration, static field
>>    declaration, or static initializer that encloses the class instance
>>    creation expression is not S, then a compile-time error occurs.*
>>
>> Should there be an *Otherwise* at the beginning of the newly added
>> sentence? (And same thing in the previous paragraph regarding anonymous
>> classes)
>>
>> -Archie
>>
>> * See JDK-8345953 <https://bugs.openjdk.org/browse/JDK-8345953> JEP 492:
>> instantiating local classes in a different static context should not be
>> allowed
>>
>> --
>> Archie L. Cobbs
>>
> ​
>


-- 
Archie L. Cobbs
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/amber-spec-experts/attachments/20241210/82176501/attachment.htm>