RE: gnubol: How do we parse this language, anyway?

[Mike Mckernan <mckernan@const.zko.dec.com>]

>>>>> "Bill" == "William M. Klein" <wmklein@ix.netcom.com>
>>>>> wrote the following on Fri, 3 Dec 1999 08:55:36 -0600

  Bill> All of this discussion seems to think that given If A = B add
  Bill> a to b size error add c to d NOT ON SIZE ERROR Display "abc"
  Bill> .

  Bill> there is some "existing code base" that does treat NOT ON
  Bill> SIZE ERROR as matching the 1st not the 2nd ADD.  The *only*
  Bill> major commercial compiler that I know of that allows this as
  Bill> an extension (Micro Focus) most definitely matches it with
  Bill> the 2nd Add.  The IBM mainframe compilers all give serious
  Bill> compiler errors for it. Fujtisu rejects it. (I don't
  Bill> personally know Unisys or Dec - but I believe they reject it
  Bill> as well.)

I also doubt that any known implementation compiles unterminated
conditionals the way Bob describes.  You arrived on the list just a
day or two too late to have seen the results of our (much too small to
be statistically significant) experiment.  I think you may be
underestimating the number of "permissive" compilers out there.

Two compilers, HP 3000 COBOL 85 and Compaq/digital COBOL 85 rejected
unterminated conditionals with severe errors.

Three compilers, MicroFocus on alpha/tru64, Fujitsu on MS Windows and
Wang ReSource either on HP/UX or AIX accepted such programs without
warnings.  None were tested with FIPS flagging set.

We had hoped that someone would be able to try this program on one or
more IBM compilers, but that has not happened, at least not yet.

I offered a disclaimer about the Wang compiler, because I was pretty
deeply involved in its development, and I probably would have been in
favor of being permissive about this, but I can't remember exactly how
the decision was made.

You don't know me, so there's no reason to be influenced by my
opinion, but we did have a bright young man on the Wang team who
seemed inordinately concerned about standards and conformance.  Let's
see, what was his name... Oh yeah, Don Shricker.

Best regards,

Mike




--
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.