Re: [PATCH] PR 78534 Change character length from int to size_t

[Andre Vehreschild <vehre at gmx dot de>]

Hi Janne,

I found that you are favoring build_int_cst (size_type_node, 0) over
size_zero_node. Is there a reason to this?

Furthermore did I have to patch this:

diff --git a/gcc/fortran/dump-parse-tree.c b/gcc/fortran/dump-parse-tree.c
index 585f25d..f374558 100644
--- a/gcc/fortran/dump-parse-tree.c
+++ b/gcc/fortran/dump-parse-tree.c
@@ -465,7 +465,7 @@ show_expr (gfc_expr *p)
          break;
 
        case BT_HOLLERITH:
-         fprintf (dumpfile, "%dH", p->representation.length);
+         fprintf (dumpfile, "%zdH", p->representation.length);
          c = p->representation.string;
          for (i = 0; i < p->representation.length; i++, c++)
            {

to bootstrap on x86_64-linux/f23.

And I have this regression:

FAIL: gfortran.dg/allocate_deferred_char_scalar_1.f03   -O1  (test for excess
errors)

allocate_deferred_char_scalar_1.f03:184:0:

     p = '12345679'
 
Warning: '__builtin_memcpy' writing 8 bytes into a region of size 5 overflows
     the destination [-Wstringop-overflow=]
     allocate_deferred_char_scalar_1.f03:242:0:

     p = 4_'12345679'
 
Warning: '__builtin_memcpy' writing 32 bytes into a region of size 20 overflows
the destination [-Wstringop-overflow=]

I also tried with a sanitized gfortran and am seeing some issues there. I have
to sort through these, but thought to let you know about the above already.

Btw, the patch for changing the ABI of the coarray-libs is already nearly done.
I just need to figure that what the state of regressions is with and without my
change.

- Andre
-- 
Andre Vehreschild * Email: vehre ad gmx dot de