Re: java and the interactive subsystem

[Steve Richter <stephenrichter@xxxxxxxxx>]

On 8/22/05, Jon Paris <Jon.Paris@xxxxxxxxxxxxxx> wrote:
>  >> But if you make calls to certain Java library methods you will end up
> multithreading even though you don't explicitly multithread your main
> application, I believe.
> 
> But multithreading per se is not the issue.  The issue only arises when a
> non-thread-safe process (e.g. those associated with the workstation I/O) is
> called from a threaded process under circumstances where it is possible that
> a second thread could also be trying to invoke the same process.
> Multi-threaded languages like Java can be run quite safely from within
> interactive jobs or (for example) the Java HSSF classes could not be used
> from an interactive job - which they can.
> 
> If pure multi-threading were the issue, Steve would also have problems with
> his proposal to call RPG code since RPG code's run-time at the current
> release is not thread safe.  It offers the ability to force threads to wait
> until the previous thread has completed processing, but that brings its own
> problems (e.g. deadly embrace).
> 
> It seems to me that as long as your design ensures that the non-thread safe
> functions are only ever called by a single active thread then you shouldn't
> have a problem.
> 
> I'm going to copy Bruce Vining on this to see if he has a more definitive
> answer.

I am very interested in reading what Bruce Vining has to say on the
subject.  ( whereever it is posted ) Java, as a managed code language,
is just too good not to be used for green screen programming.  The
fact it is not must be because of limitiations in the interactive
subsystem.

-Steve