We didn't get the "convert COBOL to JAVA", then "call a COBOL subprog from
the JAVA code" concept to work.  We had to combine the main COBOL program
(the one that was to be converted into JAVA), and the sub COBOL program into
one single COBOL program, and convert that whole program to one JAVA
program.

It's important to note that, as of yet, we have not got the multi-thread
issue resolved (we simply can't get multiple threads to work correctly), and
if we can't get that resolved, we will end up having to scrap the COBOL to
JAVA project, and opt for one of several other methods.

R.S.


        -----Original Message-----
        From:   Lars Appel [SMTP:[log in to unmask]]
        Sent:   Saturday, March 03, 2001 3:14 PM
        To:     [log in to unmask]
        Subject:        Re: Java and MR

        >Actually, that was how the 'system' was going to operate in the
beginning, a
        >JAVA prog was to call a COBOL prog.  However, we are not JAVA
folks, recall
        >that we have a product that converts COBOL to JAVA (it needs some
tweaking).

        Ah. Some tool that converts COBOL to JAVA. I missed that part in the
        original posting. I'm still not sure about the "was to call a COBOL
        subprogram" part. When calling native code subroutines from JAVA,
you
        still might want to make sure that the subroutines don't have
problems
        living in the multithreaded context that JAVA establishes. I seem to
        recall that certain types of COBOL subprograms (or prodecures in
Pascal
        speak) retain values of their local variables across invocations.
This
        could cause ugly effects if two or more JAVA threads try to use
them.

        Lars.


        PS: When mentioning my "using Java as gateway to native MPE
programs"
        experiments in my previous posting, I also forgot the URL to
examples:
        http://www.editcorp.com/Personal/Lars_Appel/JavaDemo ->
Telnet/Proglets
        (note that this is all somewhat experimental code, but worked for
me)