[Info-vax] Java portability and VMS

[Arne Vajhøj]

On 18-02-2011 10:12, hb wrote:
> On Feb 18, 4:24 am, Wendell<wendel... at yahoo.com>  wrote:
>> Of course, it always takes special care to write Java programs to be
>> platform-independent, but this sounds like any significant application
>> would require special porting to run on VMS. Is that true?
>
> If you look at this from another viewpoint you have to admit that Java
> itself is not portable. Java runs only on one platform: the JVM, the
> Java Virtual Machine.

That has always been Stroustrups point.

And it is correct, but it really does not mean anything.

> Here that is, how good the JVM support is on VMS. For example, whether
> the JVM's expectations on a filesystem were mapped to the ODS2/5 file
> system and vice versa.

Java on VMS provides several options to control that.

>                  Or whether the JVM's system calls, for example
> fork, could be implemented on VMS.

Java exposes a thread model not a fork model.

>                                   And, hopefully mentioned in the
> cited article, how many resources the JVM on VMS needs: there can be
> annoying problems caused by limited quotas. And finally, all these
> adjustments to VMS may use even more resources, than you expect,
> resulting in slow or unexpected performance even on big, fast VMS
> systems.

Trying to run a 2010 program on 2000 hardware usually result
in a resource problem.

But yes - new Java software are written for hardware where
memory is bought as 2 GB modules. On commodity servers that
costs like 100 USD.

Arne