Possible bug in javac 8 'legacy' type inference

[Liam Miller-Cushon]

Thanks Vicente,

If it ends up being a javac 7 bug, this might still be a good thing to
document for SE 8.

Liam


On Fri, Nov 15, 2013 at 7:34 AM, Vicente-Arturo Romero-Zaldivar <
vicente.romero at oracle.com> wrote:

>  Hi Liam,
>
> Thanks for your report, this needs more research but in any case I think
> that if the legacy type inference in javac 8 doesn't yield the same result
> as javac 7 we have a bug in javac 7 or in javac 8 simulation of javac 7
> type inference. So I have created bug entry [1] to track this issue.
>
> Vicente.
>
> [1] https://bugs.openjdk.java.net/browse/JDK-8028442
>
>
> On 14/11/13 17:31, Liam Miller-Cushon wrote:
>
>  The following program compiles with javac 7, and with javac 8 using
> -source 8.
>
>  However, when compiled with javac 8 and -source 7 an inference error is
> reported:
>
>  incompatible types: inference variable T#1 has incompatible upper bounds
> Message<T#2>,Result
>     Result result = getThing(key);
>                             ^
>   where T#1,T#2 are type-variables:
>     T#1 extends Message<T#1> declared in method <T#1>getThing(Key)
>     T#2 extends Result
>
>
>  Is this a bug, or is javac 8 more accurately reflecting the intent of JLS
> 15.12.2.7/15.12.2.8?
>
> ===
>
> class Message<T extends Message<T>> {}
> class Key extends Message<Key> {}
> class Result extends Message<Result> {}
>
>  abstract class Foo {
>   abstract <T extends Message<T>> T getThing(Key key);
>
>    void test(Key key) {
>     Result result = getThing(key);
>   }
> }
>
>  ===
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/compiler-dev/attachments/20131115/901ac780/attachment-0001.html