Re: Mainframe Interactive Programming

On Fri, 2018-06-08 at 07:55 -0500, Joe Pluta wrote:

On 6/7/2018 2:52 PM, dlclark@xxxxxxxxxxxxxxxx wrote:

So, for psuedo-conversational programming (as it's called), you
simply keep all the important stuff in your own communication area (which
is in-memory) and this maintains state. Of course, you can also keep
(hidden) things on the screen (as you can on the IBM i) but in the 3270
world this takes up screen real estate. So, not recommended. But, you
can keep all screen-type information in a temp storage queue so that it is
available across screen interactions, too.

I just had a serious 40-year flashback to NEP-MRT programming on the
System/3 model 15D. Get a session ID from the screen and use that to
restore all your session variables.

Here's a picture of one: http://www.ibmsystem3.nl/stories/Jenny_C.html.
I am now officially old.

Not quite as old... but I vaguely remember doing that on the S36. But,
I'll be honest, apart from remembering the term - NEP-MRT - I remember
exactly ziltch of how it was done. I also recall having to break up a
larger program on occasion and, if I recall correctly, using the display
file - or possibly the LDA - to pass key(ish) information between so it
looked like one single program when the original one blew past the
maximum size. I honestly can't remember what the maximum size was, but
it was a right PITA when it happened because invariably the huge
monolithic original program wasn't designed to be cut up and every
single line had at least one indicator denoting which display record was
currently being processed. Did the language even have subroutines, it
was such a long time ago, or was I just unlucky that the original
programmer hadn't thought to use them?