RFR: Make Zero build and run with JDK8

[Christian Thalinger]

On Oct 12, 2012, at 5:22 AM, Roman Kennke <rkennke at redhat.com> wrote:

> Am Freitag, den 12.10.2012, 07:23 -0400 schrieb Coleen Phillmore:
>> Hi,
>> I looked at this briefly and it looks okay from the cpp interpreter 
>> part.  Thank you for getting zero working again.   We tried to keep it 
>> in sync with the permgen removal changes but didn't regularly test 
>> this.   Hello world as a test for this is probably enough to verify your 
>> changes to the non-JSR292 part.
>> For JSR292 testing, you need to get a full jdk repository and go to the 
>> jdk/test/java/lang/invoke and run the tests with jtreg.   How to run and 
>> install jtreg is on the openjdk pages, and it's happily easy to do.
> 
> Ah, and before you mention it as well, I downloaded and run JRuby, the
> examples that ship with JRuby run just as well with Zero as they do with
> x86 hotspot. However, I am not sure if this exercises JSR292 or not, or
> what would be necessary to do so, etc.

If you're running on 8 it uses invokedynamic (JRuby disabled invokedynamic for 7).  For 7 you can turn it on with -Xcompile.invokedynamic=true (note the +indy in the version string):

$ jruby -v -e ""
jruby 1.7.0.RC2 (1.9.3p203) 2012-10-12 3b90805 on Java HotSpot(TM) Server VM 1.8.0-ea-b60 +indy [SunOS-x86]
$ jruby -v 
jruby 1.7.0.RC2 (1.9.3p203) 2012-10-12 3b90805 on Java HotSpot(TM) Server VM 1.7.0_10-ea-b11 [SunOS-x86]
$ jruby -v -Xcompile.invokedynamic=true
jruby 1.7.0.RC2 (1.9.3p203) 2012-10-12 3b90805 on Java HotSpot(TM) Server VM 1.7.0_10-ea-b11 +indy [SunOS-x86]

-- Chris

> 
> Roman
> 
>