See comments below for some of my "responses". I thought I was clear that I
was NOT talking about "performance" or "storage" impacts. I think most of
those types of "impacts" are cases where you and I disagree.
"Steve Comstock" <[EMAIL PROTECTED]> wrote in message
news:<442C6C2B.7020105
<snip>
> > - DYNAM (only if CANCEL is used)
> This impacts runtime behavior whether or not CANCEL is used
DYNAM and NODYNAM do not impact the output/results of a program UNLESS a
CANCEL is used. CANCEL of a program CALLed via a literal with NODYNAM does
NOT place the subprogram in "initial state" - while it does with DYNAM
specified. What other run-time behavior difference do you know of?
>
> > - FASTSRT (only if DECLARATIVES are used)
> This impacts runtime behavior whether or not DECLARATIVES are used,
> I believe
I know of NO time that FASTSRT impacts run-time results (other than the
OBVIOUS performance implication) except for when a DECLARATIVE should be
"reached" by a "bad SORT". This works with NOFASTSRT, but not with FASTSRT.
> > - PGMNAME
> I don't see the runtime behavior impact of this
Depending upon the setting of PGMNAME, a CALL to "A2345678" may or may not
find a subprogram with the name "a2345678" or "A234567899"
>
> > - QUOTE/APOST (figurative constant, not literals)
> I don't see the runtime behavior impact of this
With QUOTE the figurative constant QUOTE is
"
With APOST it is
'
That impacts comparisons and often DISPLAY and written output
> In some sense, TEST can impact runtime behavior.
I think that IBM accepts as "APARABLE" and case where a programs that are
compiled with various TEST options and the programs run to completion give
different results. Do you know of cases where semantics ARE impacted?
>
> ARITH can impact runtime behavior
You are probably right. I was thinking that most of the differences would
be detected at compile-time if you tried using numeric items over 18 digits.
However, this can impact "Intermediate results" - even if all data
individual items are limited to 18 digits.
>
> DLL can impact runtime behavior
How?
>
> OPTIMIZE can impact runtime behavior
Again, like "TEST", I believe that IBM accepts as APARable any differences
in the results (other than performance) from various settings of OPT. Even
the "removal" of non-referenced data items can't impact run-time BEHAVIOR
for "valid" COBOL programs.
>
> OUTDD can impact runtime behavior
I thought about including this. However, I think that LE will do a
DYNAMIC allocation of whatever this points to, so that the results are
always the same - just different DDNames.
>
> RENT can impact runtime behavior (along with DATA)
How? (Other than where storage is allocated? This impacts performance
but not RESULTS for two different programs that both run to successful
conclusion.
>
> In some sense, RMODE can impact runtime behavior
>
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
