(j3.2006) [Third thoughts about DIM arguments]

[Bill Long]

Malcolm Cohen wrote:
> Robert Corbett wrote:
>> As I noted
>> in my second e-mail, we need to decide whether or not "absent" means
>> the same thing as "not present,"
> 
> We already decided (and so did English).
> 
>> and whatever we decide needs to be
>> applied consistently in the specifications of the intrinsic functions.
> 
> I agree that consistent wording is good.  The standard is full of wording 
> inconsistencies though.  We (i.e. not just me) have been trying to straighten 
> out some of the most egregious ones, but ones that are stylistic differences 
> have not been considered so important at this time.
> 
>> I shall suggest that if we decide "absent" maens "not present," we
>> say "not present" instead of "absent."

I see merit in this idea in that  'present' is the same word as the name 
of  the PRESENT intrinsic, making it very clear.

> 
> I resist this suggestion.  "absent" and "present" are fine antonyms.  All of the 
> occurrences of "absent" in the standard seem to be ok to me - they are all about 
> optional dummy arguments.  I think we caught all the (many) uses of "present" 
> that were not about optional dummy arguments and changed them (usually to 
> "appears", sometimes to "is specified", etc.).  We didn't look for "omitted" - 
> in hindsight, an omission.
> 
> However, in the context of optional dummy arguments, "omitted" absolutely has to 
> mean "not present".  It would be a serious violation of the design principles of 
> optional dummy arguments to draw a distinction between textually not present and 
> textually present but runtimely not present.  I agree that "omitted" is a poor 
> choice of word though since it is more often used to mean "does not appear" 
> syntactically.

"Omitted" appears many times in the standard, especially in the I/O 
section (omitted specifiers) and seems to  always mean "does not 
appear".  There is an instance in 12 where we talk about the 
passed-object dummy argument being omitted, but in that case at [292:8] 
it means "does not appear". It looks like we're consistent in having 
"omitted" mean "does not appear".

Cheers,
Bill


-- 
Bill Long                                           longb at cray.com
Fortran Technical Support    &                 voice: 651-605-9024
Bioinformatics Software Development            fax:   651-605-9142
Cray Inc./Cray Plaza, Suite 210/380 Jackson St./St. Paul, MN 55101