On Feb 11, 2011, at 1:02 AM, Rémi Forax wrote:

> On 02/11/2011 12:20 AM, Attila Szegedi wrote:
>> I was actually doing this until yesterday for compiling the dynalink 
>> framework, and while the 1.6 javac did give me a warning and suggested I 
>> need a newer version of the compiler, it didn't stop with an error. When 
>> those classes were then loaded in an OpenJDK JVM, they nicely passed all 
>> tests. So funnily enough, I was able to work on it with the 1.6 compiler :-)
>> 
>> Of course, in my code, I never have actual invokedynamic call sites in the 
>> bytecode, it's all about direct MethodHandle manipulation.
>> 
>> Attila.
>> 
> 
> You are not alone.
> I don't want to denounce someone but I know a ruby guy ( :) ) that also 
> generate 1.6 classfile.

:-)

FWIW, I was doing it on purpose, as I (mistakenly) thought it'd be required to 
use dynalink in conjunction with the JSR-292 backport on a Java 6 JDK. One 
unexpected side benefit this had was that I was still able to measure test code 
coverage with EMMA…

Attila.

> 
> Rémi
_______________________________________________
mlvm-dev mailing list
[email protected]
http://mail.openjdk.java.net/mailman/listinfo/mlvm-dev

Reply via email to