RE: java.awt.headless rears it's ugly head

I am leaving the JVM version as shipped by not overriding properties in
a systemdefault.properties file, so an iteration of the java properties
class reveals a raft of entries like -

java.vm.version=1.4
java.runtime.version=1.4.2_16-b05
java.class.version=48.0
os.version=V5R4M0
java.specification.version=1.4
java.vm.specification.version=1.0
java.specification.vendor=Sun Microsystems Inc.
java.vm.info=build JDK-1.4, native threads, jitc_de
java.version=1.4.2

Since java.awt.headless=true presently breaks existing production code
then I have resorted to offloading the request to a server job via a
data queue.

Peter

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx
[mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Scott Klement
Sent: Friday, 5 June 2009 9:46 a.m.
To: RPG programming on the IBM i / System i
Subject: Re: java.awt.headless rears it's ugly head

hi Peter,

Thanks for doing the research. I had not realized that this only
happens when you use the native AWT libraries.

Good to know.

What version of the JVM are you testing this with?


Peter Connell wrote:

Scott,

I can only comment on the issue as it presented itself to me. I had

not

encountered it before.
The RPG/java app had been in test for a week when testers reported
failure of other existing code.
I narrowed it down as follows.

1. Remove /qibm/userdata/java400/systemDefault.properties.
2. call qp2shell '/QOpensys/bin/ls'
3. Run any RPG that will create a JVM by running a simple class.
4. call qp2shell '/QOpensys/bin/ls'

All is well.

Now create /qibm/userdata/java400/systemDefault.properties
Add an entry for java.awt.headless=true or os400.awt.native=true
OR
Load the property via ADDENVVAR ENVVAR(QIBM_RPG_JAVA_PROPERTIES)
VALUE('-Djava.awt.headless=true;')

Repeat steps 2+ above and the last call to qp2shell will throw CPFB9C7

Peter

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx
[mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Scott Klement
Sent: Friday, 5 June 2009 4:09 a.m.
To: RPG programming on the IBM i / System i
Subject: Re: java.awt.headless rears it's ugly head

Hi Peter,

Peter Connell wrote:

Everything works fine until a totally unrelated task runs the

QP2SHELL

program subsequently in the same job which immediately throws the

error

CPFB9C7 - i5/OS PASE already running in this job.

Shouldn't this happen with any Java code running via RPG? You make

it

sound like this only happens if you use the headless property, but I
don't think that's true.

The issue is that only one copy of PASE can be active at a time. Java

runs in PASE, and therefore any Java code at all (irrespective of the
headless property) should cause this to fail.

Or am I mistaken?