RE: Various IBM extensions (was RE: gnubol: Re: FW: RCF - SC26-9046-03 - LRM (Special-Names format error)

["William M. Klein" <wmklein@ix.netcom.com>]

> -----Original Message-----
>
>
> You know, a lot of these extensions are probably bugs that IBM
> then found people
> had come to depend on.
>
> Tim Josling
>

I don't want to get too "off-track", but where OS/VS COBOL is (was) probably
one of the MOST common COBOL compilers ever in use, I don't find this too
irrelevant.

At one time, I headed up the "COBOL Project" at GUIDE (one of the 2 IBM
mainframe user groups).  At that time, migration from the old ('74 and '68)
OS/VS COBOL compiler to the newer VS COBOL II (and then COBOL/370 and then
COBOL for this-and-that) was one of the major issues that we talked about.

The history of the OS/VS COBOL compiler is a long (and "checkered") one which
basically boiled down to the fact that IBM had started with COBOL-F or
earlier (circa 1965) and kept "servicing" it and never really reviewed what
it did.  The final result when they DID do a rewrite (which is what VS COBOL
II is) was that no-one at IBM or any where else really knew what OS/VS COBOL
allowed.  In included such truly "obscure" things as DISPLAY-ST which
provided support for English monetary system PRE-decimalization (pounds,
shilling, pence, etc - even though the UK had dropped this system years
before); Report-Writer syntax that bore almost NO relationship to the ANSI
Standard (IBM never did upgrade that from the '68 Standard to the '74
Standard); and a HUGE number of "syntax" kludges that lead to the rumor that
with a little luck, you could actually feed a PL/I source program into the
compiler and get "an executable load module out of it". <G>

One of the interesting journeys into "What's in a word?" that we went thru
was working with IBM to come up with an accurate (and legally acceptable to
IBM lawyers) term for this "stuff".  We went thru:

  "undocumented extension"
  "undetected user-error"
  "bugs in the compiler"

and a variety of other things - all of which simply meant that the OS/VS
COBOL compiler accepted whatever it accepted (and that tended to differ
depending on maintenance level) - and UNLESS a feature were documented in the
LRM, then IBM would provide migration assistance when and if it pleased
them - and not in other cases.

The relevance to the GNU project is that there really is a lot of OS/VS COBOL
and/or originally OS/VS COBOL code out there (in the real world) - but that
trying to base compiler development based on compatibility with that is
PROBABLY not what you want to do (and certainly isn't what IBM did).

Bill Klein
  wmklein <at> ix.netcom.com


--
This message was sent through the gnu-cobol mailing list.  To remove yourself
from this mailing list, send a message to majordomo@lusars.net with the
words "unsubscribe gnu-cobol" in the message body.  For more information on
the GNU COBOL project, send mail to gnu-cobol-owner@lusars.net.