procyberian/gcc
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

re PR fortran/24783 ([4.1 and 4.2 only] Implicit none in module overwrite explicit in procedure)

Browse source 
fortran/ChangeLog
2006-11-20  Bernhard Fischer  <aldot@gcc.gnu.org>

        PR fortran/24783
        * resolve.c (resolve_variable): Get the implicit type from the
        symbols namespace rather than the default namespace. Fix whitespace.
        (resolve_formal_arglist, resolve_equivalence): Fix typo.


testsuite/ChangeLog
2006-11-20  Bernhard Fischer  <aldot@gcc.gnu.org>

        PR fortran/24783
        * gfortran.dg/implicit_10.f90: New test.

From-SVN: r119016
This commit is contained in:
Bernhard Fischer 2006-11-20 17:20:33 +01:00 • committed by Bernhard Reutner-Fischer
parent d58542ea03
commit 9d691ba750
4 changed files with 48 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
gcc/fortran/ChangeLog
View file

@ -1,3 +1,10 @@
2006-11-20 Bernhard Fischer <aldot@gcc.gnu.org>
PR fortran/24783
* resolve.c (resolve_variable): Get the implicit type from the
symbols namespace rather than the default namespace. Fix whitespace.
(resolve_formal_arglist, resolve_equivalence): Fix typo.
2006-11-19 Erik Edelmann <eedelman@gcc.gnu.org>
* resolve.c (resolve_ref): Check for ALLOCATABLEs to the right of

8
gcc/fortran/resolve.c
View file

@ -232,7 +232,7 @@ resolve_formal_arglist (gfc_symbol * proc)
{
gfc_error
("Character-valued argument '%s' of statement function at "
"%L must has constant length",
"%L must have constant length",
sym->name, &sym->declared_at);
continue;
}
@ -2966,7 +2966,7 @@ resolve_variable (gfc_expr * e)
else
{
/* Must be a simple variable reference. */
if (gfc_set_default_type (sym, 1, NULL) == FAILURE)
if (gfc_set_default_type (sym, 1, sym->ns) == FAILURE)
return FAILURE;
e->ts = sym->ts;
}
@ -6008,11 +6008,9 @@ resolve_symbol (gfc_symbol * sym)
case FL_PARAMETER:
if (resolve_fl_parameter (sym) == FAILURE)
return;
break;
default:
break;
}
@ -6692,7 +6690,7 @@ resolve_equivalence (gfc_equiv *eq)
{
if (value_name != NULL)
{
gfc_error ("Initialized objects '%s' and '%s' cannot both "
gfc_error ("Initialized objects '%s' and '%s' cannot both "
"be in the EQUIVALENCE statement at %L",
value_name, sym->name, &e->where);
continue;

5
gcc/testsuite/ChangeLog
View file

@ -1,3 +1,8 @@
2006-11-20 Bernhard Fischer <aldot@gcc.gnu.org>
PR fortran/24783
* gfortran.dg/implicit_10.f90: New test.
2006-11-19 Gabriel Dos Reis <gdr@integrable-solutions.net>
PR c++/8586

33
gcc/testsuite/gfortran.dg/implicit_10.f90 Normal file
View file

@ -0,0 +1,33 @@
! { dg-do run }
! Check fix for PR24783 where we did try to resolve the implicit type
! from the wrong namespace thus rejecting valid code.
MODULE mod1
IMPLICIT NONE
CONTAINS
SUBROUTINE sub(vec, ny)
IMPLICIT REAL (a-h,o-z)
IMPLICIT INTEGER (i-n)
DIMENSION vec(ny)
ny = fun(vec(ny),1,1)
RETURN
END SUBROUTINE sub
REAL FUNCTION fun(r1, i1, i2)
IMPLICIT REAL (r,f)
IMPLICIT INTEGER (i)
DIMENSION r1(i1:i2)
r1(i1) = i1 + 1
r1(i2) = i2 + 1
fun = r1(i1) + r1(i2)
END FUNCTION fun
END MODULE mod1
use mod1
IMPLICIT REAL (d)
INTEGER i
dimension di(5)
i = 1
if (fun(di(i),1,2).NE.5) call abort()
call sub(di(i),i)
if (i.NE.4) call abort()
end
! { dg-final { cleanup-modules "mod1" } }