Re: CFINT: I understand it now...

["Leif Svalgaard" <leif@xxxxxxxx>]

From: Paul Raulerson <praulerson@hot.rr.com>
> I don't agree with this chain of thought. In fact, server applications
> when well designed, are very efficient. Interactive programs can be
> very efficient, but in an AS/400 environment, they are nowhere near
> as efficient as say, CICS applications. (Even given that AS/400
> interactive applications are probably more efficient that anything
> else out there except for old Wang VS programs.... ;)
>

Paul,
This is very perceptive. CICS is the granddaddy of C/S applications.
Green screens served by a single (well, modern CICS can have many)
server process. A great deal of the efficiency of pseudo-conversational
CICS came from the very many restrictions that the programmer had
to observe (no Getmain, minimal 'context' saved from one transaction
to the next, no use of STRING/UNSTRING, etc). These restrictions
were actually good as they imposed a strict discipline on the programmer
having to split up a large program into several 'transactions' (screen
exchanges). Thirty years ago a CICS system with 128K memory
could efficiently support hundreds of 'interactive' users on a machine
100 times as slow as today's machines. We'll never get back there
again. The damage has been done. Bloat rules.

Although there are (were?) third party CISC providers on the AS/400,
they never caught on because the S/38 was designed to provide
superior interactive capacity without having to use the transactional
model of CICS. The Pluta approach is a return to CICS. CICS lives!