[J3] DIM arguments

[Bill Long]

Ok, thanks for clarifying.   This is definitely how the standard is written - even the comment about it in the Introduction is correct.  The implementation is now easy - just remove the error message that used to be issued for supplying an optional argument as the actual corresponding to DIM. 

Cheers,
Bill



> On Mar 6, 2019, at 6:35 PM, Malcolm Cohen via J3 <j3 at mailman.j3-fortran.org> wrote:
> 
> Bill, it is exactly like an ordinary generic overload as has been writable by the user since 1991.
>  
> There is no issue here.  If DIM textually appears, it is required to be present, exactly the same as in ANY procedure call where the corresponding dummy argument is not OPTIONAL. 
>  
> Of course it’s invalid to do ANY(A,DIM) when DIM is absent.  It’s also invalid to do ABS(DIM) or DIM=3 or X=DIM+1 etc etc etc etc etc.
>  
> > the program just gets a segfault at runtime
>  
> At least some compilers have an option to error terminate with an explanatory message when something invalid is done with an absent optional argument.  You’re not *REQUIRED* to just seg fault.
>  
> Cheers,
> -- 
> ..............Malcolm Cohen, NAG Oxford/Tokyo.
>  
> From: J3 <j3-bounces at mailman.j3-fortran.org> On Behalf Of Bill Long via J3
> Sent: Thursday, March 7, 2019 9:23 AM
> To: General J3 interest list <j3 at mailman.j3-fortran.org>
> Cc: Bill Long <longb at cray.com>; Malcolm Cohen <malcolm at nag-j.co.jp>
> Subject: Re: [J3] DIM arguments
>  
> I agree that’s what the standard says. However, the challenge comes with this version:
> 
> subroutine boo(a,i)
> logical,intent(in) :: a(10,10)
> integer,optional :: i
> 
> print *, any(a,i)
> 
> end
> 
> I you call this with
> 
> CALL BOO (A, 1)
> 
> then the code is valid. If called with 
> 
> CALL BOO(A)
> 
> it is invalid now that the DIM argument to ANY is no longer optional. So the intention is that the compiler always generates the two-argument version, and if the user (thinking the I argument to BOO is optional) calls BOO with only one argument, the program just gets a segfault at runtime? We can do that, but I thought this was the issue F08/0038 was trying to address. 
> 
> Cheers,
> Bill
> 
> 
> 
> > On Mar 6, 2019, at 6:04 PM, Malcolm Cohen via J3 <j3 at mailman.j3-fortran.org> wrote:
> > 
> > In F2018, the DIM argument of ANY is not an optional dummy argument. There are two forms of ANY: one with no DIM argument, one with a non-OPTIONAL DIM argument.
> > 
> > Thus this follows the usual rules. Viz, in 15.5.2.12 Argument presence and restrictions on arguments not present, it says “An optional dummy argument that is not present is subject to the following restrictions… It shall not be supplied as an actual argument corresponding to a nonoptional dummy argument…”
> > 
> > Thus the code is valid, and the reference is to the second form of ANY, not the first. I repeat: the DIM argument of ANY is not OPTIONAL, it is just like a generic overload.
> > 
> > Cheers,
> > -- 
> > ..............Malcolm Cohen, NAG Oxford/Tokyo.
> > 
> > From: J3 <j3-bounces at mailman.j3-fortran.org> On Behalf Of Bill Long via J3
> > Sent: Thursday, March 7, 2019 8:21 AM
> > To: General J3 interest list <j3 at mailman.j3-fortran.org>
> > Cc: Bill Long <longb at cray.com>
> > Subject: [J3] DIM arguments
> > 
> > For this example:
> > 
> > subroutine boo(a,i)
> > logical,intent(in) :: a(10,10)
> > integer,optional :: i
> > if (present(i)) then
> > print *, any(a,i)
> > else
> > print *, any(a)
> > end if
> > end
> > 
> > F08 gave an error because there was a specific requirement in the (then) optional DIM argument: 
> > 
> > ""The corresponding actual argument shall not be an optional dummy argument.”
> > 
> > In F2018 we changed the DIM argument to be not optional. But we also deleted the restriction sentence. Was that intentional? It seems that (according to the logic in interp F08/0038) that we did not want to allow optional arguments to correspond to the DIM argument of the ANY function. But the restrictions on the use of optional arguments alone (absent the specific prohibition for the DIM argument of ANY) appear to make the above code conforming. Or maybe the prohibition is stated elsewhere for these DIM arguments, and I’m not seeing it. 
> > 
> > Cheers,
> > Bill
> > 
> > 
> > Bill Long longb at cray.com
> > Principal Engineer, Fortran Technical Support & voice: 651-605-9024
> > Bioinformatics Software Development fax: 651-605-9143
> > Cray Inc./ 2131 Lindau Lane/ Suite 1000/ Bloomington, MN 55425
> > 
> > 
> > Disclaimer
> > The Numerical Algorithms Group Ltd is a company registered in England and Wales with company number 1249803. The registered office is: Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom. Please see our Privacy Notice for information on how we process personal data and for details of how to stop or limit communications from us.
> > 
> > This e-mail has been scanned for all viruses and malware, and may have been automatically archived by Mimecast Ltd, an innovator in Software as a Service (SaaS) for business.
> 
> Bill Long longb at cray.com
> Principal Engineer, Fortran Technical Support & voice: 651-605-9024
> Bioinformatics Software Development fax: 651-605-9143
> Cray Inc./ 2131 Lindau Lane/ Suite 1000/ Bloomington, MN 55425

Bill Long                                                                       longb at cray.com
Principal Engineer, Fortran Technical Support &   voice:  651-605-9024
Bioinformatics Software Development                      fax:  651-605-9143
Cray Inc./ 2131 Lindau Lane/  Suite 1000/  Bloomington, MN  55425