| Chuck Stevens 2004-09-06, 3:55 am |
| Xref: newsfeed-west.nntpserver.com comp.lang.cobol:94814
"LX-i" <lxi0007@netscape.net> wrote in message
news:NnvYc.34591$5s3.5528@fe40.usenetserver.com...
> Granted, that's not quite the way he said it, but I think that was his
point.
May be, but that's exactly the issue. I suspect it is probably true that a
*majority* of implementations store fixed-point numeric items as integers
and keep track of the necessary scale factors in the compiler-generated
code, and that may indeed be what he meant. I don't believe the standard
*requires* this, but I think a preference toward integer arithmetic is
implicit in the COBOL standards.
I might even be persuaded to agree with him if he had amended "All
fixed-point numbers are integers"
to be something like "Most COBOL environments store and process fixed-point
numeric values as integers" early on in the thread.
What his point (his *conclusion*) was is irrelevant when the *premise* he
drew upon in order to arrive at that point is faulty. Where he tends to
lose credibility is not in making the error in the first place, it is in the
continued instance that, as stated, it was correct and everybody disagreeing
was wrong. In this case, it is not only the original statement but the
*continued insistance* that all fixed-point numbers *really are* integers,
that is absurd.
-Chuck Stevens
|