Re: ClassPath Woes - What the Heck is happening.

[Barbara Morris <bmorris@xxxxxxxxxx>]

"Bartell, Aaron L. (TC)" wrote:
> 
> Tried that, then it replaces anything else that might be in the CLASSPATH
> variable.  I can't guarantee that I am the only one using the CLASSPATH.
> It's just easier to say "Here, add this to the classpath, if it's already
> there shoot me an error and I will handle it".  One could write some
> routines to see if a particular entry exists in the CLASSPATH but I just
> found this to be easier.
> 
> Are there any gotcha's with this approach?
> 

In V5R2, you can only start the JVM once.  (In V5R1, you can usually
start it many times, but it's not really supported so you occasionally
can only get the 1.1 JDK on restarts.  I wouldn't build a dependency in
V5R1 on being able to bring down the JVM and restart it, given this
restriction in V5R2.  See "Java Programming Changes" in the V5R2 Memo to
Users http://publib.boulder.ibm.com/iseries/v5r2/ic2924/info/rzaq9.pdf.)

So I think the CLASSPATH should be setup once, at the start of the job,
with all the directories for any application that might need Java.  It
might even be good to have a *SYS CLASSPATH envvar, so it would be added
automatically to every job when the job starts.