https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70070
kargl at gcc dot gnu.org changed:
What |Removed |Added
----------------------------------------------------------------------------
Priority|P3 |P4
--- Comment #10 from kargl at gcc dot gnu.org ---
(In reply to kargl from comment #9)
> Programs z1.f90, z2.f90, z3.f90, z5.f90 now give the error
>
> troutmask:sgk[271] gfcx -o z z3.f90 && ./z
> a.f90:4:9:
>
> 4 | data (c(i:i), i=1,999) /999*'c'/
> | 1
> Error: Invalid substring in data-implied-do at (1) in DATA statement
>
> Oddly, za.f90 compiles because the variable is implicitly defined
> to be of type integer. If 'integer i' is added to this case, then
> an error is issued.
This was actually fixed by
------------------------------------------------------------------------
r257962 | kargl | 2018-02-24 09:22:10 -0800 (Sat, 24 Feb 2018) | 11 lines
2018-02-24 Steven G. Kargl <ka...@gcc.gnu.org>
PR fortran/30792
* decl.c (gfc_match_data): Check for invalid substring in
data-implied-do
2018-02-24 Steven G. Kargl <ka...@gcc.gnu.org>
PR fortran/30792
* gfortran.dg/data_substring.f90: New test.
but the check for issuing the error message enforced that the
implied-do index must be integer. For za.f90 testcase, the
basic type for the implicitly typed 'i' is BT_UNKNOWN. Checking
for BT_INTEGER is tto strict. Here a patch against svn r 280157.
Whoever decides to commit the patch should probably convert
za.f90 to a testcase.
Index: gcc/fortran/decl.c
===================================================================
--- gcc/fortran/decl.c (revision 280157)
+++ gcc/fortran/decl.c (working copy)
@@ -657,7 +657,6 @@ gfc_match_data (void)
goto cleanup;
if (new_data->var->iter.var
- && new_data->var->iter.var->ts.type == BT_INTEGER
&& new_data->var->iter.var->symtree->n.sym->attr.implied_index == 1
&& new_data->var->list
&& new_data->var->list->expr
