[J3] an amusing mistake

[Bill Long]

> On Aug 4, 2020, at 9:58 AM, Robert Corbett via J3 <j3 at mailman.j3-fortran.org> wrote:
> 
> Oops.  After fixing my mistake,
> the second program violates C725, which states
> 
> C725 (R722) The optional comma
> in a /length-selector/ is permitted
> only if no double-colon
> separator appears in the
> /type-declaration-stmt/.

The extra comma did seem very weird.  I never see programs use that in practice, and am unclear why it is even allowed. I assume it is a relic of fixed-source form, which is Obsolescent. (Of course, CHARACTER* is also obsolescent, but I do still see that in very old codes.)  Indeed, C725 is in tiny (obsolescent) font. 


> The type-declaration-stmt in the
> second program contains a
> double-colon separator, and so
> the optional comma should not
> appear.
> 
> I have no doubt that the intent was
> to ban the optional comma only if
> the double-colon precedes the
> /entity-decl-list/, but that is not
> what the constraint says.

Perhaps a more reasonable solution would be to just move the optional comma to the Deleted List. (Hence deleting the constraint). 


Cheers,
Bill


> 
> Bob Corbett
> 
>> On Aug 4, 2020, at 6:27 AM, Bill Long via J3 <j3 at mailman.j3-fortran.org> wrote:
>> 
>> 
>> 
>>> On Aug 4, 2020, at 6:29 AM, Robert Corbett via J3 <j3 at mailman.j3-fortran.org> wrote:
>>> 
>>> In the past few days, I had reasons to read
>>> the Fortran 2018 standard more closely than
>>> I usually read it.  As often happens when I
>>> read the standard closely, I saw problems
>>> that I overlooked before.  Yesterday, I saw
>>> a problem I found amusing.  (Yes, I am
>>> easily amused.)
>>> 
>>> Consider the two programs
>>> 
>>> -------------------------------------------
>>>     PROGRAM ONE
>>>       CHARACTER*1, C(SUM([1]))
>>>     END
>>> -------------------------------------------
>>>     PROGRAM TWO
>>>       CHARACTER*1, C(SUM([INTEGER:: 1])
>>>     END
>>> -------------------------------------------
>>> 
>>> The first program is standard conforming.
>>> The second program is not a conforming
>>> program under a strict reading of the
>>> standard.
>> 
>> Or not so strict.   It is not conforming because of a missing right paren at the end of the second statement.
>> 
>> With that syntax error fixed, both compile for me.
>> 
>> Cheers,
>> Bill
>> 
>> 
>> 
>>> Both programs should be
>>> standard conforming.
>>> 
>>> I suspect there are compilers that 
>>> accept both programs and other compilers
>>> that reject both programs.  I doubt there
>>> are any compilers that accept the first
>>> program and reject the second.
>>> 
>>> Bob Corbett
>>> 
>> 
>> Bill Long                                                                       longb at hpe.com
>> Engineer/Master , Fortran Technical Support &   voice:  651-605-9024
>> Bioinformatics Software Development                      fax:  651-605-9143
>> Hewlett Packard Enterprise/ 2131 Lindau Lane/  Suite 1000/  Bloomington, MN  55425
>> 
>> 
>> 
>> 
> 

Bill Long                                                                       longb at hpe.com
Engineer/Master , Fortran Technical Support &   voice:  651-605-9024
Bioinformatics Software Development                      fax:  651-605-9143
Hewlett Packard Enterprise/ 2131 Lindau Lane/  Suite 1000/  Bloomington, MN  55425