Is IBM java 1.5 dangerous for scientific production?

Hi,

I want to post a warning about IBM Java 1.5 for use in scientific research 
and education. I have been working with a scientific software company 
named Chemaxon. They are famous for developing tools and Java libraries to 
manipulate and treat chemical informations. They require Java Sun 1.5. I 
did try their products using IBM Java 1.5 as distributed by Novell in my 
SLED10SP1. The softwares seemed to work with the IBM JVM. But, the 
resulting treatments were dependent on the precise JVM in use. The problem 
is that no warning is ever emitted and only careful examination of the 
results and comparison with a Sun JVM may lead to detection of errors 
induced by the use of unsupported JVM.

I believe the problem not specific to Chemaxon. This is of course 
extremely dangerous in projects where Java is involved and where 
computational results are used for decision.

Does somebody know why Novell is distributing only one version of the JVM 
1.5? And why is it the IBM one? It is not so common otherwise?

In my opinion, access to the largest possible choice of JVM is a priority 
to guarantee the compatibility of key Java applications with the one 
chosen by the developers.

Regards,
Gilles Marcou

If you want to discuss the specific Java verson included with SLES 10, I 
suggest you post in one of the SLES forums.
I think the reason for Novell using the IBM JVM by default is a support 
cost issue. While Sun distributes Java for free to customers, companies 
that want professional support for java versions included with their 
products pay a lot. My guess is that Novell gets a much better deal from 
IBM for the same level of support. In any case, even if the IBM JVM is 
installed by default, nothing prevents you from using Sun's JVM is you 
need it for compatibility with your application.

-- 
Marcel Cox (Sysop)