(j3.2006) Constraint C835: SELECT TYPE selector must be a named variable

[Bader, Reinhold]

Actually, upon re-reading 16.5.1.6 of 10-007r1, I would feel more confident about what I said if that
section contained a sentence like

If the selector is allocatable, it shall be allocated and shall not be deallocated during execution of the block;

instead of simply

If the selector is allocatable, it shall be allocated;

(and similar for the POINTER case). Perhaps a glitch that needs fixing in the standard?

Cheers
Reinhold

> -----Urspr?ngliche Nachricht-----
> Von: j3-bounces at mailman.j3-fortran.org [mailto:j3-bounces at mailman.j3-
> fortran.org] Im Auftrag von Damian Rouson
> Gesendet: Donnerstag, 6. November 2014 22:20
> An: fortran standards email list for J3
> Betreff: Re: (j3.2006) Constraint C835: SELECT TYPE selector must be a named
> variable
> 
> Thanks!  That makes sense. I suspected it had to do with stripping certain
> properties from the selector, but I hadn't thought about the ugly consequences.
> 
> It would be great if there were some language to explain what is meant by
> "named variable".  It took me a bit of time to figure out why the offending
> example wasn't acceptable.
> 
> ________________________________
> Damian Rouson, Ph.D., P.E.
> Founder & President, Sourcery, Inc.
> 510-600-2992 (mobile)
> http://www.sourceryinstitute.org
> http://rouson.youcanbook.me
> 
> 
> 
> On Nov 6, 2014, at 1:03 PM, Bader, Reinhold <Reinhold.Bader at lrz.de> wrote:
> 
> > Hello Damian,
> >
> > I've occasionally been wondering about this as well. The answer I've
> > come up with privately is that the association has the property of
> > removing the ALLOCATABLE attribute. Enforcing the use of an associate
> > name in the scenario you prescribe then precludes rather ugly
> > scenarios like
> >
> > SELECT TYPE(bar%stuff)
> > TYPE IS (foo)
> >  deallocate(bar%stuff)
> >  allocate(type_from_which_foo_is_extended :: bar%stuff)
> > bar%stuff%component_which_does_not_exist_anymore = ...
> > ...
> > END SELECT
> >
> > Cheers
> >
> > Reinhold
> >
> >
> >> -----Urspr?ngliche Nachricht-----
> >> Von: j3-bounces at mailman.j3-fortran.org [mailto:j3-bounces at mailman.j3-
> >> fortran.org] Im Auftrag von Damian Rouson
> >> Gesendet: Donnerstag, 6. November 2014 21:54
> >> An: fortran standards email list for J3
> >> Betreff: (j3.2006) Constraint C835: SELECT TYPE selector must be a
> >> named variable
> >>
> >> Could someone tell me the motivation for the following constraint on
> >> "select type" selectors in the Fortran 2008 standard?
> >>
> >> C835 (R847) If selector is not a named variable, associate-name => shall
> appear.
> >>
> >> I'm guessing this constraint explains the Cray compiler error message below
> >> (gfortran gives essentially the same error message).   What's the motivation
> for
> >> disallowing the second "select type" statement below?
> >>
> >> On a possibly related note, Note 6.1 states that "P%AGE" is a
> >> "variable".  How does a "variable" differ from a "named variable"?  I
> >> can find the phrase "named variable" in only a few places in the standard, but
> I can't find a definition of it.
> >>
> >> Damian
> >>
> >>
> >> p01923 at swan:~> cat selecttype.f90
> >> program main
> >> implicit none
> >> type object
> >> end type
> >> type foo
> >>  class(object), allocatable :: stuff
> >> end type
> >> type(foo) :: bar
> >>
> >> select type(barstuff=>bar%stuff ) ! This compiles cleanly end select
> >>
> >> select type(bar%stuff ) ! This gives an error end select
> >>
> >> end
> >> p01923 at swan:~> ftn selecttype.f90
> >>
> >> select type(bar%stuff ) ! This gives an error
> >>            ^
> >> ftn-1875 crayftn: ERROR MAIN, File = selecttype.f90, Line = 13,
> >> Column = 15 The SELECT TYPE selector must be a named variable or an
> >> associate-name must appear.
> >>
> >> Cray Fortran : Version 8.3.4
> (u83056f83186i83169p83291a83009e83011z83291)
> >> Cray Fortran :               (x8318r83015w83011t8311b83037)
> >> Cray Fortran : Sun Nov 02, 2014  10:08:52 Cray Fortran : Compile
> >> time:  0.0040 seconds Cray Fortran : 16 source lines Cray Fortran : 1
> >> errors, 0 warnings, 0 other messages, 0 ansi Cray Fortran : "explain
> >> ftn-message number" gives more information about each message.
> >> _______________________________________________
> >> J3 mailing list
> >> J3 at mailman.j3-fortran.org
> >> http://mailman.j3-fortran.org/mailman/listinfo/j3
> > _______________________________________________
> > J3 mailing list
> > J3 at mailman.j3-fortran.org
> > http://mailman.j3-fortran.org/mailman/listinfo/j3
> 
> _______________________________________________
> J3 mailing list
> J3 at mailman.j3-fortran.org
> http://mailman.j3-fortran.org/mailman/listinfo/j3