[Patch, Fortran, comitted] Fix test-suite failure due to REAL(10)

[Tobias Burnus]

Tobias Burnus wrote:
> array_constructor_23.f contained REAL(10) which failed on all platforms
> which have not such a type such as PowerPC which has REAL(16).
>
> I created a REAL(8) version, which was probably not really needed.
>
> Checked in as Rev. 134678.
>   
Same for GCC 4.3.0, except I did not include the real(8) version.

Rev. 134715

Tobias


Index: ChangeLog
===================================================================
--- ChangeLog   (Revision 134714)
+++ ChangeLog   (Arbeitskopie)
@@ -1,3 +1,7 @@
+2008-04-26  Tobias Burnus  <burnus@net-b.de>
+
+       * gfortran/array_constructor_23.f: Change REAL(10) into kind > 8.
+
 2008-04-26  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

        PR fortran/35994
Index: gfortran.dg/array_constructor_23.f
===================================================================
--- gfortran.dg/array_constructor_23.f  (Revision 134714)
+++ gfortran.dg/array_constructor_23.f  (Arbeitskopie)
@@ -1,4 +1,5 @@
 ! { dg-do run }
+! { dg-require-effective-target fortran_large_real }
 ! Tests the fix for PR35944/6/7, in which the variable array constructors below
 ! were incorrectly translated and wrong code was produced.
 !
@@ -34,7 +35,7 @@
       subroutine fa2083
       implicit none
       integer j1,k
-      parameter (k=10)              !failed
+      parameter (k=selected_real_kind (precision (0.0_8) + 1)) ! failed
       REAL(k) QDA1(10)
       REAL(k) QDA(10), qval
       qda = (/ 1,2,3,4,5,6,7,8,9,10 /)