[Bug fortran/84276] Invalid error for valid statement function

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

kargl at gcc dot gnu.org changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution|--- |FIXED
   Target Milestone|--- |6.5

--- Comment #18 from kargl at gcc dot gnu.org ---
Fixed on 6-branch, 7-branch, and trunk.

[Bug fortran/84276] Invalid error for valid statement function

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #17 from kargl at gcc dot gnu.org ---
Author: kargl
Date: Mon Feb 12 18:25:41 2018
New Revision: 257596

URL: https://gcc.gnu.org/viewcvs?rev=257596=gcc=rev
Log:
2018-02-12  Steven G. Kargl  

PR fortran/54223
PR fortran/84276
* interface.c (compare_actual_formal): Add in_statement_function
bool parameter.  Skip check of INTENT attribute for statement
functions.  Arguments to a statement function cannot be optional,
issue error for missing argument.
(gfc_procedure_use, gfc_ppc_use, gfc_arglist_matches_symbol): Use
in_statement_function.

2018-02-12  Steven G. Kargl  

PR fortran/54223
PR fortran/84276
* gfortran.dg/statement_function_1.f90: New test.
* gfortran.dg/statement_function_2.f90: New test.

Added:
branches/gcc-6-branch/gcc/testsuite/gfortran.dg/statement_function_1.f90
branches/gcc-6-branch/gcc/testsuite/gfortran.dg/statement_function_2.f90
Modified:
branches/gcc-6-branch/gcc/fortran/ChangeLog
branches/gcc-6-branch/gcc/fortran/interface.c
branches/gcc-6-branch/gcc/testsuite/ChangeLog

[Bug fortran/84276] Invalid error for valid statement function

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #16 from kargl at gcc dot gnu.org ---
Author: kargl
Date: Mon Feb 12 18:04:33 2018
New Revision: 257593

URL: https://gcc.gnu.org/viewcvs?rev=257593=gcc=rev
Log:
2018-02-12  Steven G. Kargl  

PR fortran/54223
PR fortran/84276
* interface.c (compare_actual_formal): Add in_statement_function
bool parameter.  Skip check of INTENT attribute for statement
functions.  Arguments to a statement function cannot be optional,
issue error for missing argument.
(gfc_procedure_use, gfc_ppc_use, gfc_arglist_matches_symbol): Use
in_statement_function.

2018-02-12  Steven G. Kargl  

PR fortran/54223
PR fortran/84276
* gfortran.dg/statement_function_1.f90: New test.
* gfortran.dg/statement_function_2.f90: New test.

Added:
branches/gcc-7-branch/gcc/testsuite/gfortran.dg/statement_function_1.f90
branches/gcc-7-branch/gcc/testsuite/gfortran.dg/statement_function_2.f90
Modified:
branches/gcc-7-branch/gcc/fortran/ChangeLog
branches/gcc-7-branch/gcc/fortran/interface.c
branches/gcc-7-branch/gcc/testsuite/ChangeLog

[Bug fortran/84276] Invalid error for valid statement function

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #15 from kargl at gcc dot gnu.org ---
Author: kargl
Date: Sun Feb 11 18:44:05 2018
New Revision: 257565

URL: https://gcc.gnu.org/viewcvs?rev=257565=gcc=rev
Log:
2018-02-11  Steven G. Kargl  

PR fortran/54223
PR fortran/84276
* interface.c (compare_actual_formal): Add in_statement_function
bool parameter.  Skip check of INTENT attribute for statement
functions.  Arguments to a statement function cannot be optional,
issue error for missing argument.
(gfc_procedure_use, gfc_ppc_use, gfc_arglist_matches_symbol): Use
in_statement_function.

2018-02-11  Steven G. Kargl  

PR fortran/54223
PR fortran/84276
* gfortran.dg/statement_function_1.f90: New test.
* gfortran.dg/statement_function_2.f90: New test.

Added:
trunk/gcc/testsuite/gfortran.dg/statement_function_1.f90
trunk/gcc/testsuite/gfortran.dg/statement_function_2.f90
Modified:
trunk/gcc/fortran/ChangeLog
trunk/gcc/fortran/interface.c
trunk/gcc/testsuite/ChangeLog

[Bug fortran/84276] Invalid error for valid statement function

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #14 from Steve Kargl  ---
On Sat, Feb 10, 2018 at 02:20:20AM +, mecej4 at outlook dot com wrote:
> 
> Will keyword arguments in statement function references be retained as a GNU
> extension?
> 

The extension will remain, but I intend to hope a new bug
report about the issue.

[Bug fortran/84276] Invalid error for valid statement function

[mecej4 at outlook dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #13 from mecej4  ---
Thanks for locating the relevant clauses. With your extracts serving as a
roadmap, I can find essentially the same wording in F2003, but with different
numbers for rules and constraints, in section 12.4.

R1217 function-reference is procedure-designator ( [ actual-arg-spec-list ] )

C1220 (R1217) The procedure-designator shall designate a function.

R1220 actual-arg-spec is [ keyword = ] actual-arg

C1225 (R1220) The keyword = shall not appear if the interface of the procedure
is implicit in the 9 scoping unit.

It struck me as odd that, since statement functions were marked "obsolescent"
even in F95, references to them would be given features that were not available
in F77, such as keyword arguments.

Will keyword arguments in statement function references be retained as a GNU
extension?

[Bug fortran/84276] Invalid error for valid statement function

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #12 from Steve Kargl  ---
On Fri, Feb 09, 2018 at 09:15:00PM +, mecej4 at outlook dot com wrote:
> > 
> > I cannot find a prohibition in F2018 standard.
> > AFAICT, gfortran always supported keywords, and
> > I've developed a patch to retain that behavior.
> 
> I cannot find an explicit prohibition, either, but please consider these
> related statements from F2003:
> 

I found where the prohibition occurs in F95, F2003, and F2018.
If you want to get a headache, read F2018, 15.5.  It's a little
convoluted to chase down all the terms because of the wonderful
OOP language, but one has

R1520 function-reference  is procedure-designator ( [ actual-arg-spec-list ] )

C1523 (R1520) The procedure-designator shall designate a function.
...
R1523 actual-arg-spec  is [ keyword = ] actual-arg
...
C1530 (R1523) The 'keyword =' shall not appear if the interface of
the procedure is implicit.

The first thing to chase down is 'procedure' in C1530 means 'function
or subroutine'.  Then one needs to chase down that 'function' means
'function (in a Fortran subprogram sense) or statement function.'

[Bug fortran/84276] Invalid error for valid statement function

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #11 from Steve Kargl  ---
On Fri, Feb 09, 2018 at 09:15:00PM +, mecej4 at outlook dot com wrote:
> --- Comment #10 from mecej4  ---
> (In reply to Steve Kargl from comment #9)
> > On Fri, Feb 09, 2018 at 01:52:03PM +, mecej4 at outlook dot com wrote:
> > > (In reply to Steve Kargl from comment #7)
> > > > ...
> > > >   w = qofs(i = i, s = hh)
> > > > ...
> > > 
> > > Keyword argument lists are not allowed for statement
> > > functions, are they? I realize that the compiler will
> > > have to detect and flag attempts to use keyword
> > > arguments as disallowed, but no code generation is needed.
> > 
> > I cannot find a prohibition in F2018 standard.
> > AFAICT, gfortran always supported keywords, and
> > I've developed a patch to retain that behavior.
> 
> I cannot find an explicit prohibition, either, but please consider
> these related statements from F2003:
> 
> Section 12.3.1: ..."The interface of a statement function is always
> implicit".

Yep, I found that.

> Section 12.3.1.1: ..."A procedure other than a statement function
> shall have an explicit interface if it is referenced and a reference
> to the procedure appears with an argument keyword"

Yep, I found that, too.  This contains requirements for procedures
that aren't statement functions.  As a statement functions is an
exception and a compiler builds its interface at compile time,
I do not see why the standard would prohibit the use of keywords.
Although I don't recall ever seeing a bug report about keywords
and statement functions, over the years, I have come to expect that
users can find new and unique ways to break gfortran.  My patch is
keeping gfortran's historical behavior.  A side benefit is that an
additional line of code fixes PR 54223.

Unfortunately, getting expert input on c.l.f has become too much
drama.

[Bug fortran/84276] Invalid error for valid statement function

[mecej4 at outlook dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #10 from mecej4  ---
(In reply to Steve Kargl from comment #9)
> On Fri, Feb 09, 2018 at 01:52:03PM +, mecej4 at outlook dot com wrote:
> > (In reply to Steve Kargl from comment #7)
> > > ...
> > >   w = qofs(i = i, s = hh)
> > > ...
> > 
> > Keyword argument lists are not allowed for statement functions, are they? I
> > realize that the compiler will have to detect and flag attempts to use 
> > keyword
> > arguments as disallowed, but no code generation is needed.
> > 
> 
> I cannot find a prohibition in F2018 standard.
> AFAICT, gfortran always supported keywords, and
> I've developed a patch to retain that behavior.

I cannot find an explicit prohibition, either, but please consider these
related statements from F2003:

Section 12.3.1: ..."The interface of a statement function is always
implicit".

Section 12.3.1.1: ..."A procedure other than a statement function shall
have an explicit interface if it is referenced and a reference to the procedure
appears with an argument keyword"

That still leaves the prohibition unstated. However, I find the following in
the Fortran 2003 Handbook (Adams et al), p. 451:

14. A statement function is referenced in the same manner as any other
function, except that because statement function interfaces are implicit, the
keyword form of actual arguments is not allowed; the argument association rules
are the same.

[Bug fortran/84276] Invalid error for valid statement function

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #9 from Steve Kargl  ---
On Fri, Feb 09, 2018 at 01:52:03PM +, mecej4 at outlook dot com wrote:
> (In reply to Steve Kargl from comment #7)
> > Ugh.  Statement functions should be removed from the Standard.
> > The simply fix, of course, does not work if someone is clever
> > and uses keywords in a reference that involves a statement
> > function.
> > 
> > ...
> >   w = qofs(i = i, s = hh)
> > ...
> 
> Keyword argument lists are not allowed for statement functions, are they? I
> realize that the compiler will have to detect and flag attempts to use keyword
> arguments as disallowed, but no code generation is needed.
> 

I cannot find a prohibition in F2018 standard.
AFAICT, gfortran always supported keywords, and
I've developed a patch to retain that behavior.

[Bug fortran/84276] Invalid error for valid statement function

[mecej4 at outlook dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

mecej4  changed:

   What|Removed |Added

 CC||mecej4 at outlook dot com

--- Comment #8 from mecej4  ---
(In reply to Steve Kargl from comment #7)
> Ugh.  Statement functions should be removed from the Standard.
> The simply fix, of course, does not work if someone is clever
> and uses keywords in a reference that involves a statement
> function.
> 
> ...
>   w = qofs(i = i, s = hh)
> ...

Keyword argument lists are not allowed for statement functions, are they? I
realize that the compiler will have to detect and flag attempts to use keyword
arguments as disallowed, but no code generation is needed.

[Bug fortran/84276] Invalid error for valid statement function

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #7 from Steve Kargl  ---
Ugh.  Statement functions should be removed from the Standard.
The simply fix, of course, does not work if someone is clever
and uses keywords in a reference that involves a statement
function.

  subroutine step(hh, h, s, w)
  real, intent(inout) :: h, hh, s
  real, intent(out) :: w
  real :: qofs
  integer i
  qofs(s, i) = i * s
  i = 42
  w = qofs(hh, i)
  w = qofs(i = i, s = hh)
  end subroutine step

[Bug fortran/84276] Invalid error for valid statement function

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #6 from Steve Kargl  ---
On Thu, Feb 08, 2018 at 06:53:00PM +, kargl at gcc dot gnu.org wrote:
>
> I have a patch.
> 

The patch is incomplete.  If the actual and dummy arguments
type and type parameter match then, everything works ok.
If there is a mismatch, then we get an ICE.  I'm working
on it.

  subroutine stepns(hh,h,s,w)
  real, intent(inout) :: h,hh,s
  real, intent(out)  :: w
  real :: qofs
  integer i
  qofs(s)=s
  w=qofs(hh+h)  ! OK
  i = 42
  w = qofs(i)   ! Causes ICE, but should issue error.
  end subroutine stepns

[Bug fortran/84276] Invalid error for valid statement function

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

kargl at gcc dot gnu.org changed:

   What|Removed |Added

   Assignee|unassigned at gcc dot gnu.org  |kargl at gcc dot gnu.org

--- Comment #5 from kargl at gcc dot gnu.org ---
I have a patch.

[Bug fortran/84276] Invalid error for valid statement function

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

Dominique d'Humieres  changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed||2018-02-08
 Ever confirmed|0   |1
  Known to fail||4.3.1

--- Comment #4 from Dominique d'Humieres  ---
Confirmed.

[Bug fortran/84276] Invalid error for valid statement function

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

Richard Biener  changed:

   What|Removed |Added

   Keywords||rejects-valid
 Status|NEW |UNCONFIRMED
   Last reconfirmed|2018-02-08 00:00:00 |
 Ever confirmed|1   |0

[Bug fortran/84276] Invalid error for valid statement function

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

Dominique d'Humieres  changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed||2018-02-08
 Ever confirmed|0   |1

--- Comment #3 from Dominique d'Humieres  ---
Confirmed.

[Bug fortran/84276] Invalid error for valid statement function

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #2 from Steve Kargl  ---
On Thu, Feb 08, 2018 at 02:26:50AM +, sgk at troutmask dot
apl.washington.edu wrote:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276
> 
> --- Comment #1 from Steve Kargl  ---
> Reduced testcase.
> 
>   subroutine stepns(hh,h,s,w)
>   real, intent(inout) :: h,hh,s
>   real, intent(out)  :: w
>   real :: qofs
>   qofs(s)=s
>   w=qofs(hh+h)
>   end subroutine stepns
> 
> Problem occurs with interface.c:3219.
> 

So, the problem appears to boil down to gfortran acquiring
not only the type and type parameters from the declaration
statements, but also the attributes.  Statement functions
should not inherit the attributes.

[Bug fortran/84276] Invalid error for valid statement function

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

--- Comment #1 from Steve Kargl  ---
Reduced testcase.

  subroutine stepns(hh,h,s,w)
  real, intent(inout) :: h,hh,s
  real, intent(out)  :: w
  real :: qofs
  qofs(s)=s
  w=qofs(hh+h)
  end subroutine stepns

Problem occurs with interface.c:3219.

[Bug fortran/84276] Invalid error for valid statement function

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84276

kargl at gcc dot gnu.org changed:

   What|Removed |Added

   Priority|P3  |P4
  Known to fail||6.4.1, 7.3.1, 8.0