[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #3 from janus at gcc dot gnu.org 2011-02-03 20:45:04 UTC ---
Test case:


type :: t
  integer :: i
end type

class(t), allocatable :: x
type(t) :: y = t (3)

x = y
print *,x%i
end

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #4 from janus at gcc dot gnu.org 2011-02-03 20:56:10 UTC ---
Here is a simple patch for getting rid of the error message:


Index: gcc/fortran/resolve.c
===================================================================
--- gcc/fortran/resolve.c       (revision 169520)
+++ gcc/fortran/resolve.c       (working copy)
@@ -8879,15 +8879,13 @@ resolve_ordinary_assign (gfc_code *code, gfc_names
        gfc_current_ns->proc_name->attr.implicit_pure = 0;
     }

-  /* F03:7.4.1.2.  */
-  /* FIXME: Valid in Fortran 2008, unless the LHS is both polymorphic
-     and coindexed; cf. F2008, 7.2.1.2 and PR 43366.  */
-  if (lhs->ts.type == BT_CLASS)
-    {
-      gfc_error ("Variable must not be polymorphic in assignment at %L",
-                &lhs->where);
-      return false;
-    }
+  /* F03:7.4.1.2, F08:7.2.1.2.  */
+  /* Valid in Fortran 2008, unless the LHS is both polymorphic and coindexed. 
*/
+  if (lhs->ts.type == BT_CLASS
+      && gfc_notify_std (GFC_STD_F2008,
+                        "Fortran 2008: Variable must not be polymorphic in "
+                        "assignment at %L", &lhs->where) == FAILURE)
+    return false;

   /* F2008, Section 7.2.1.2.  */
   if (gfc_is_coindexed (lhs) && gfc_has_ultimate_allocatable (lhs))

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

janus at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |rejects-valid
                 CC|                            |pault at gcc dot gnu.org

--- Comment #5 from janus at gcc dot gnu.org 2011-02-03 21:04:47 UTC ---
Realloc-on-assign for scalars was implemented in r169356.

Paul, what does it take to make this work for polymorphic scalars?

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #6 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-02-03 21:24:38 UTC ---
(In reply to comment #4)
> Here is a simple patch for getting rid of the error message:

The patch is not quite right. The LHS must be allocatable
("gfc_expr_attr(e).allocatable") - and it must not be coindexed
("!gfc_is_conindexed(e)").

Thus, you need something like:

if (lhs->ts.type == BT_CLASS)
  {
    if (gfc_expr_attr(e).allocatable") && !gfc_is_conindexed(e))

      if (gfc_notify_std(F2008 ...)  == FAILURE)
         return false;
    else
      {
        gfc_error ();
        return false;
      }
  }

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #7 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-02-03 21:28:10 UTC ---
(In reply to comment #4)
> Here is a simple patch for getting rid of the error message:

The patch is not quite right. The LHS must be allocatable and it must not be
coindexed nor a coarray - otherwise, an intrinsic allocate is not allowed.

Cf. "7.2.1.2 Intrinsic assignment statement": "(1) if the variable is
polymorphic it shall be allocatable and not a coarray," + quote in comment 0.

Thus, you need something like:

if (lhs->ts.type == BT_CLASS)
  {
    if (gfc_expr_attr(e).allocatable") && !gfc_is_coindexed(e)
        && !gfc_expr_attr(e).codimension)
      {
        if (gfc_notify_std(F2008 ...)  == FAILURE)
           return false;
      }
    else
      {
        gfc_error ();
        return false;
      }
  }

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #8 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-02-03 21:29:12 UTC ---
(In reply to comment #6)
Stupid firefox - it somehow must have submitted a draft. Ignore comment 6 and
look at comment 7.

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[pault at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #9 from Paul Thomas <pault at gcc dot gnu.org> 2011-02-04 09:32:01 UTC ---
(In reply to comment #5)
> Realloc-on-assign for scalars was implemented in r169356.
> 
> Paul, what does it take to make this work for polymorphic scalars?

Not a lot, I think.  If you take a look in gfc_trans_allocate, there is a bit
that I think that you must have written, which initialises CLASS variables.  I
would have thought that this can be lifted and deposited wholesale in
trans_assignment1 and gfc_trans_allocate reconfigured to use the assignement.

I am taking a look at gfc_trans_allocate this lunchtime; there's something not
quite right structurally with it.  I attempted to add a patch of code to have a
single evaluation of the SOURCE expression and wound up displacing unrelated
declarations, when I added the se.post to the end of the main block. gimple did
not like it at all :-(  I'll come back to you.

Cheers

Paul

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

janus at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
         AssignedTo|unassigned at gcc dot       |janus at gcc dot gnu.org
                   |gnu.org                     |

--- Comment #10 from janus at gcc dot gnu.org 2011-02-10 22:41:47 UTC ---
The following patch makes the test case in comment #3 work correctly:


Index: gcc/fortran/trans-expr.c
===================================================================
--- gcc/fortran/trans-expr.c    (revision 169986)
+++ gcc/fortran/trans-expr.c    (working copy)
@@ -5838,6 +5838,14 @@ is_scalar_reallocatable_lhs (gfc_expr *expr)
     && !expr->ref)
     return true;

+  /* An allocatable class variable with no reference.  */
+  if (expr->symtree->n.sym->ts.type == BT_CLASS
+      && CLASS_DATA (expr->symtree->n.sym)->attr.allocatable
+      && expr->ref && expr->ref->type == REF_COMPONENT
+      && strcmp (expr->ref->u.c.component->name, "_data") == 0
+      && expr->ref->next == NULL)
+    return true;
+
   /* All that can be left are allocatable components.  */
   if ((expr->symtree->n.sym->ts.type != BT_DERIVED
     && expr->symtree->n.sym->ts.type != BT_CLASS)
Index: gcc/fortran/resolve.c
===================================================================
--- gcc/fortran/resolve.c    (revision 169987)
+++ gcc/fortran/resolve.c    (working copy)
@@ -8881,14 +8881,32 @@ resolve_ordinary_assign (gfc_code *code, gfc_names
     gfc_current_ns->proc_name->attr.implicit_pure = 0;
     }

-  /* F03:7.4.1.2.  */
-  /* FIXME: Valid in Fortran 2008, unless the LHS is both polymorphic
-     and coindexed; cf. F2008, 7.2.1.2 and PR 43366.  */
   if (lhs->ts.type == BT_CLASS)
     {
-      gfc_error ("Variable must not be polymorphic in assignment at %L",
-         &lhs->where);
-      return false;
+      /* F08:7.2.1.2.  */
+      if (!gfc_expr_attr (lhs).allocatable)
+    {
+      gfc_error ("Polymorphic variable in intrinsinc assignment must be "
+             "allocatable at %L", &lhs->where);
+      return false;
+    }
+      else if (gfc_expr_attr (lhs).codimension)
+    {
+      gfc_error ("Polymorphic variable in intrinsinc assignment must not be"
+             " coarray at %L", &lhs->where);
+      return false;
+    }
+      else if (gfc_is_coindexed (lhs))
+    {
+      gfc_error ("Polymorphic variable in intrinsinc assignment must not be"
+             " coindexed at %L", &lhs->where);
+      return false;
+    }
+      /* F03:7.4.1.2.  */
+      else if (gfc_notify_std (GFC_STD_F2008, "Fortran 2008: Variable must not
"
+                   "be polymorphic in assignment at %L",
+                   &lhs->where) == FAILURE)
+    return false;
     }

   /* F2008, Section 7.2.1.2.  */

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #11 from janus at gcc dot gnu.org 2011-02-10 22:56:21 UTC ---
Stupid question, but shouldn't it be possible to replace the call to
'is_scalar_reallocatable_lhs' by a check for the allocatable attribute?


Index: gcc/fortran/trans-expr.c
===================================================================
--- gcc/fortran/trans-expr.c    (revision 169986)
+++ gcc/fortran/trans-expr.c    (working copy)
@@ -6116,8 +6087,7 @@
   if (lss == gfc_ss_terminator)
     {
       /* F2003: Add the code for reallocation on assignment.  */
-      if (gfc_option.flag_realloc_lhs
-           && is_scalar_reallocatable_lhs (expr1))
+      if (gfc_option.flag_realloc_lhs && gfc_expr_attr (expr1).allocatable)
        alloc_scalar_allocatable_for_assignment (&block, rse.string_length,
                                                 expr1, expr2);


AFAICS, the routine does not do anything more anyway, right?

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #12 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-02-11 19:38:23 UTC ---
(In reply to comment #10)
> The following patch makes the test case in comment #3 work correctly:

I think your current patch only works if declared type (LHS) == effective type
RHS. I think you need to use the size of the effective types ("_size"). The
change should be in trans-expr.c's alloc_scalar_allocatable_for_assignment. Cf.
"Determine allocate size" in trans-stmt.c's gfc_trans_allocate

(In reply to comment #11)
> Stupid question, but shouldn't it be possible to replace the call to
> 'is_scalar_reallocatable_lhs' by a check for the allocatable attribute?

I guess so - though I have not studied the code.

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #13 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-07-22 12:08:28 UTC ---
Regarding the LHS (in addition to comment 6):

- It may not be coindexed ("!gfc_is_conindexed(e)")
- It may not be a coarray ("!gfc_expr_attr (e).codimension")
- It may not have a coarray component ("!gfc_expr_attr (e).coarray_comp")

Cf. F2008, 7.2.1.2, item (1) and (3).

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #14 from Tobias Burnus <burnus at gcc dot gnu.org> 2012-01-06 10:24:24 UTC ---
As litmus test: BT_CLASS version of PR 50981: Passing an
unallocated/unassociated [or absent + nonpointer/nonallocatable] BT_CLASS
variable or a unallocated/unassociated BT_CLASS component to an
nonpointer/nonallocatable argument of an ELEMENTAL procedure.

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #15 from Tobias Burnus <burnus at gcc dot gnu.org> 2012-02-09 17:48:11 UTC ---
Created attachment 26628
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=26628
Small patch for the resolve.c. It misses all the real work (trans*.c).

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[abenson at caltech dot edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

Andrew Benson <abenson at caltech dot edu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |abenson at caltech dot edu

--- Comment #16 from Andrew Benson <abenson at caltech dot edu> 2012-04-10 21:06:13 UTC ---
The following compiles with "-std=f2003" and runs successfully (using gfortran
4.8 r186145) with the "a=b" line in, but if I switch it for the "a%d=b%d" line
I get a "Variable must not be polymorphic in intrinsic assignment at (1) -
check that there is a matching specific subroutine for '=' operator" error:

module testMod
  implicit none
  public t
  type t
     integer :: i
  end type t
  type c
     class(t), allocatable :: d
  end type c
end module testMod

program testProg
  use testMod
  implicit none
  type(t) :: f
  type(c) :: a,b
  allocate(b%d,source=f)
  b%d%i=12345

  a=b      !! THIS COMPILES OK
!  a%d=b%d   !! THIS DOES NOT

  write (0,*) a%d%i
  write (0,*) b%d%i
end program testProg

Since "d" is polymorphic both should be disallowed with "-std=f2003" (and both
allowed under F2008 I think).

[Bug fortran/43366] [OOP][F08] Intrinsic assign to polymorphic variable

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

Bug 43366 depends on bug 46321, which changed state.

Bug 46321 Summary: [OOP] Polymorphic deallocation
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46321

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

[Bug fortran/43366] [OOP][F08] Intrinsic assign to polymorphic variable

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

--- Comment #17 from Tobias Burnus <burnus at gcc dot gnu.org> ---
Author: burnus
Date: Wed Sep 18 18:14:57 2013
New Revision: 202713

URL: http://gcc.gnu.org/viewcvs?rev=202713&root=gcc&view=rev
Log:
2013-09-15  Tobias Burnus  <burnus@net-b.de>

        PR fortran/43366
        * primary.c (gfc_variable_attr): Also handle codimension.
        * resolve.c (resolve_ordinary_assign): Add invalid-diagnostic
        * for
        polymorphic assignment.

2013-09-15  Tobias Burnus  <burnus@net-b.de>

        PR fortran/43366
        * gfortran.dg/class_39.f03: Update dg-error.
        * gfortran.dg/class_5.f03: Ditto.
        * gfortran.dg/class_53.f90: Ditto.
        * gfortran.dg/realloc_on_assign_20.f90: New.
        * gfortran.dg/realloc_on_assign_21.f90: New.
        * gfortran.dg/realloc_on_assign_22.f90: New.


Added:
    trunk/gcc/testsuite/gfortran.dg/realloc_on_assign_20.f90
    trunk/gcc/testsuite/gfortran.dg/realloc_on_assign_21.f90
    trunk/gcc/testsuite/gfortran.dg/realloc_on_assign_22.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/primary.c
    trunk/gcc/fortran/resolve.c
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gfortran.dg/class_39.f03
    trunk/gcc/testsuite/gfortran.dg/class_5.f03
    trunk/gcc/testsuite/gfortran.dg/class_53.f90

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[janus at gcc dot gnu dot org]

------- Comment #2 from janus at gcc dot gnu dot org  2010-08-03 11:23 -------
(In reply to comment #1)
> note that realloc on assignment must be supported to get this working properly

... which is PR35810.


-- 

janus at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  BugsThisDependsOn|                            |35810
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2010-08-03 11:23:14
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366

[Bug fortran/43366] [OOP][F2008] Intrinsic assign to polymorphic variable

[burnus at gcc dot gnu dot org]

------- Comment #1 from burnus at gcc dot gnu dot org  2010-03-14 14:01 -------
See "Variable must not be polymorphic in assignment" in resolve.c and note that
realloc on assignment must be supported to get this working properly (might
already work for derived types).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43366