javah output file naming weirdness

[Ian Rogers]

Andrew John Hughes wrote:
> 2008/4/28 Alan Bateman <Alan.Bateman at sun.com>:
>   
>> Ian Rogers wrote:
>>
>>     
>>> :
>>> it's a requirement that the Jikes RVM is built with at least Java 5.0 due
>>>       
>> to wide spread use of generics and annotations in the code base. I
>> personally use Java 6 to build Jikes RVM (on systems with neither Java 1.4.2
>> or Java 5.0 installed). The only time I've heard of a problem with javah is
>> Andrew when using Open JDK 6. We could have been fluking things before.
>>     
>>  Perhaps my suggestion that this might be an old javah vs. new javah issue
>> is a red herring but, just for pig iron, I created a class
>> org.jikesvvm.scheduler.greenthreads.VM_Process and ran javac + javah with
>> each of openjdk (7), Sun's jdk5, and Sun's jdk6. All generate
>> org_jikesvvm_scheduler_greenthreads_VM_Process.h. This has me curious as to
>> how org_jikesvvm_scheduler_greenthreads_VM_0005fProcess.h is generated in
>> your environment since you are using "Java 6". Is this Sun's jdk6?
>>
>>  -Alan.
>>
>>
>>
>>     
>
> Alan,
>
> Good to know I'm not the only one seeing the lack of an 0005f.  Ian,
> can you shed any more light on this?
>
> Thanks,
>   

My $LANG is en_GB, I've been using Sun JDK 5 and 6. At some points also 
JRockit and IBM's JDKs. For our regression machines we must use IBM's 
JDK on AIX machines. The $LANG on the main regression machines is 
en_AU.UTF-8 .

Repeating Alan's test (javah -d . -jni 
org.jikesrvm.scheduler.greenthreads.VM_Process) causes the creation of 
org_jikesrvm_scheduler_greenthreads_VM_Process.h for me - no 0005f. 
javah -version is "1.6.0_04". Andrew, does this mean your ant behaviour 
is the cause?

Thanks,
Ian
--
Third International Workshop on Implementation, Compilation, 
Optimization of Object-Oriented Languages, Programs and Systems 
(ICOOOLPS 2008)
Submissions/Notification/Conference: May 4th/May 19th/July 7th
Paphos (Cyprus) http://icoolps.loria.fr