Re: RUNJVA in CCSID 5026 results in JVA0122 rc 1?

Den 18/07/10 02.00, David Gibbs skrev:

On 7/17/2010 6:10 PM, Thorbjoern Ravn Andersen wrote:

I know, but I belive that if a given scenario is "you are running CCSID
5026, and the Classic JVM has a show stopper bug acknowledged by IBM
which is fixed in the J9 one" that might a good reason for the customer
to consider installing an extra JRE.

The fact that they have to install a JRE to use this particular feature of our product was hard enough a pill to swallow to begin with. :

Gah. Picky users. Hopefully your feature brings fantastic functionality they cannot live without :)

Question is, what code do you have that essentially locks you with the
Classic JVM?

None. But we try to minimize the number of LPP's that we require a customer to install.

At one time when the startup time of the classic jvm was too painful, I was considering if we could bring our own JVM for tiny stuff. The tiniest interpreter with almost full Java 5 functionality: jamvm - http://jamvm.sourceforge.net/ - runs reasonably well on PowerPC but I have not gotten around to test if it runs under AIX and hence PASE (oh, what a difference it would make if PASE supported Linux binaries and not only AIX).

The J9 jvm is much much faster at startup time. Hence my suggestion of switching.

I want the java app to run with the classic and J9 JVM's. It works fine in CCSID 37& 5035. CCSID 5026 is the only one we've encountered a problem with.

As it happens, I'm working around the problem by changing the CCSID to 5035 if they are running 5026, run the Java application, then switch back to 5026.

Heh. Glad you found a circumvention. Does not fix the bug but at least allows your users to work.