(j3.2006) badly worded semantics for the IMPORT statement

Robert Corbett Robert.Corbett
Wed Dec 5 03:17:13 EST 2007

Van Snyder wrote:
> Robert Corbett wrote:
>> Van Snyder wrote:
>>> Robert Corbett wrote:
>>>> I don't think J3 intended for it to be standard conforming.  From an
>>>> implementor's POV, I see no reason for the restriction.  Fortran 90
>>>> and later versions of Fortran require two or more compilation passes,
>>>> which makes the restriction unnecessary.  The restriction might be
>>>> there to enforce someone's idea of good style; otherwise, I see no
>>>> point to it.
>>> Isn't this controversy the definition of "an interp is needed?"
>> I don't see that there is a controversy.  Unless I misinterpreted
>> Bill's initial response, I agree with Bill as to what the language
>> in the standard regarding the IMPORT statement was intended to say.
>> I don't think it says what was intended, which is why I asserted
>> that it is "badly worded."  I assume it comes as no surprise to
>> anyone on the committee that the Fortran standard contains some
>> badly worded sections and some pointless restrictions.
> The part about "I don't think it says what was intended" is generally 
> regarded as a good reason for an interp request.
> Words do not fly magically into corrigenda.  They only get there via the 
> interp process.  If the standard needs clarification, the only way to do 
> that officially is via the interp process.  If an interp is requested, 
> the question will be pondered in depth and officially.  If no interp is 
> requested, the situation will almost surely stand as it is today, with a 
> tiny possibility of being repaired in a future standard just because 
> somebody reads it again and finds it unclear again.  Or maybe not.

I have no faith in the interpretation process resulting in improvements
to the wording of the standard.  For example, I recall Larry Rolison and
Dick Hendrikson, who are hardly novices when it comes to reading the
standard, filed request for interpretation pointing out that the
optional arguments to MAX and MIN are not identified as "optional" as
described in Section 13.2.  The edit required to fix that problem would
have been trivial.  The committee chose instead to say that the existing
text was clear enough as is.  As a result, the text is still broken in
the latest Fortran 2008 draft.

Bob Corbett

More information about the J3 mailing list