[Bug fortran/25396] New: Operator overloading for array-valued functions gets shape incorrectly

[Bug fortran/25396] New: Operator overloading for array-valued functions gets shape incorrectly

[jb at gcc dot gnu dot org]

Consider the following testcase:

module geometry

  implicit none

  interface operator(.cross.)
     module procedure cross
  end interface

contains

    ! Cross product between two 3d vectors.
    pure function cross(a, b)
      real, dimension(3), intent(in) :: a,b
      real, dimension(3) :: cross

      cross = (/ a(2) * b(3) - a(3) * b(2), &
           a(3) * b(1) - a(1) * b(3), &
           a(1) * b(2) - a(2) * b(1) /)
    end function cross

end module geometry

program opshape
  use geometry

  implicit none

  real :: t(3,3), a(3), b(3)

  ! Storing to an array works.
  b = t(:,2) .cross. t(:,3)

  ! However, when the result is not stored it doesn't work.

  ! Following two line are equivalent, but the first one causes an
  ! error.
  print *, size (t(:,2) .cross. t(:,3))
  print *, size (cross (t(:,2), t(:,3)))

  ! Same here.
  a = dot_product (t(:,1), t(:,2) .cross. t(:,3))
  a = dot_product (t(:,1), cross (t(:,2), t(:,3)))

end program opshape

The two dot_product and the two size invocations are equivalent, but the ones
that use the operator .cross. instead of calling the cross function directly,
cause errors:

~/src/gfortran/test/opshape% gfortran opshape.f90
 In file opshape.f90:37

  print *, size (t(:,2) .cross. t(:,3))
                              1
Error: 'array' argument of 'size' intrinsic at (1) must be an array
 In file opshape.f90:41

  a = dot_product (t(:,1), t(:,2) .cross. t(:,3))
                                        1
Error: 'vector_b' argument of 'dot_product' intrinsic at (1) must be of rank 1


-- 
           Summary: Operator overloading for array-valued functions gets
                    shape incorrectly
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: jb at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25396

[Bug fortran/25396] Operator overloading for array-valued functions gets shape incorrectly

[eedelman at gcc dot gnu dot org]

------- Comment #1 from eedelman at gcc dot gnu dot org  2005-12-27 19:26 -------
Confirmed.


-- 

eedelman at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |eedelman at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-12-27 19:26:24
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25396

[Bug fortran/25396] Operator overloading for array-valued functions gets shape incorrectly

[eedelman at gcc dot gnu dot org]

------- Comment #2 from eedelman at gcc dot gnu dot org  2005-12-29 20:01 -------
Working on a patch.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25396

[Bug fortran/25396] Operator overloading for array-valued functions gets shape incorrectly

[eedelman at gcc dot gnu dot org]

------- Comment #3 from eedelman at gcc dot gnu dot org  2005-12-29 20:57 -------
I think this simple one-liner fixes the bug:

Index: interface.c
===================================================================
--- interface.c (revision 109139)
+++ interface.c (working copy)
@@ -1718,6 +1718,7 @@ gfc_extend_expr (gfc_expr * e)
   e->value.function.actual = actual;
   e->value.function.esym = NULL;
   e->value.function.isym = NULL;
+  e->value.function.name = NULL;

   if (gfc_pure (NULL) && !gfc_pure (sym))
     {


but because of PR 22607, I can't test it properly.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25396

[Bug fortran/25396] Operator overloading for array-valued functions gets shape incorrectly

[eedelman at gcc dot gnu dot org]

------- Comment #4 from eedelman at gcc dot gnu dot org  2005-12-30 15:02 -------
Subject: Bug 25396

Author: eedelman
Date: Fri Dec 30 15:02:44 2005
New Revision: 109171

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=109171
Log:
fortran/
2005-12-30  Erik Edelmann  <eedelman@gcc.gnu.org>

        PR fortran/22607
        * trans-decl.c(gfc_get_extern_function_decl): Don't set
        DECL_IS_PURE (fndecl) = 1 for return-by-reference
        functions.

        fortran/PR 25396
        * interface.c (gfc_extend_expr): Initialize
        e->value.function.name to NULL.



testsuite/
2005-12-30  Erik Edelmann  <eedelman@gcc.gnu.org>

        PR fortran/22607
        * gfortran-dg/pure_byref_3.f90: New.

        fortran/PR 25396
        * gfortran.dg/userdef_operator_1.f90: New.


Added:
    trunk/gcc/testsuite/gfortran.dg/pure_byref_3.f90
    trunk/gcc/testsuite/gfortran.dg/userdef_operator_1.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/interface.c
    trunk/gcc/fortran/trans-decl.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25396

[Bug fortran/25396] Operator overloading for array-valued functions gets shape incorrectly

[eedelman at gcc dot gnu dot org]

------- Comment #5 from eedelman at gcc dot gnu dot org  2005-12-30 15:07 -------
Subject: Bug 25396

Author: eedelman
Date: Fri Dec 30 15:07:48 2005
New Revision: 109172

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=109172
Log:
fortran/
2005-12-30  Erik Edelmann  <eedelman@gcc.gnu.org>

        PR fortran/22607
        * trans-decl.c(gfc_get_extern_function_decl): Don't set
        DECL_IS_PURE (fndecl) = 1 for return-by-reference
        functions.

        fortran/PR 25396
        * interface.c (gfc_extend_expr): Initialize
        e->value.function.name to NULL.



testsuite/
2005-12-30  Erik Edelmann  <eedelman@gcc.gnu.org>

        PR fortran/22607
        * gfortran-dg/pure_byref_3.f90: New.

        fortran/PR 25396
        * gfortran.dg/userdef_operator_1.f90: New.


Added:
    branches/gcc-4_1-branch/gcc/testsuite/gfortran.dg/pure_byref_3.f90
    branches/gcc-4_1-branch/gcc/testsuite/gfortran.dg/userdef_operator_1.f90
Modified:
    branches/gcc-4_1-branch/gcc/fortran/ChangeLog
    branches/gcc-4_1-branch/gcc/fortran/interface.c
    branches/gcc-4_1-branch/gcc/fortran/trans-decl.c
    branches/gcc-4_1-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25396

[Bug fortran/25396] Operator overloading for array-valued functions gets shape incorrectly

[eedelman at gcc dot gnu dot org]

------- Comment #6 from eedelman at gcc dot gnu dot org  2005-12-30 15:11 -------
Subject: Bug 25396

Author: eedelman
Date: Fri Dec 30 15:11:15 2005
New Revision: 109173

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=109173
Log:
fortran/
2005-12-30  Erik Edelmann  <eedelman@gcc.gnu.org>

        PR fortran/22607
        * trans-decl.c(gfc_get_extern_function_decl): Don't set
        DECL_IS_PURE (fndecl) = 1 for return-by-reference
        functions.

        fortran/PR 25396
        * interface.c (gfc_extend_expr): Initialize
        e->value.function.name to NULL.



testsuite/
2005-12-30  Erik Edelmann  <eedelman@gcc.gnu.org>

        PR fortran/22607
        * gfortran-dg/pure_byref_3.f90: New.

        fortran/PR 25396
        * gfortran.dg/userdef_operator_1.f90: New.


Added:
    branches/gcc-4_0-branch/gcc/testsuite/gfortran.dg/pure_byref_3.f90
    branches/gcc-4_0-branch/gcc/testsuite/gfortran.dg/userdef_operator_1.f90
Modified:
    branches/gcc-4_0-branch/gcc/fortran/ChangeLog
    branches/gcc-4_0-branch/gcc/fortran/interface.c
    branches/gcc-4_0-branch/gcc/fortran/trans-decl.c
    branches/gcc-4_0-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25396

[Bug fortran/25396] Operator overloading for array-valued functions gets shape incorrectly

[tobi at gcc dot gnu dot org]

------- Comment #7 from tobi at gcc dot gnu dot org  2005-12-31 16:39 -------
Fixed. BTW, Erik, you can use svn merge to backport patches to the older
branches.  That would save you "svn add"ing them on all branches.


-- 

tobi at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tobi at gcc dot gnu dot org
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.0.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25396