https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87476
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |edlinger at gcc dot gnu.org,
| |jakub at gcc dot gnu.org,
| |jason at gcc dot gnu.org
--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I see two options out of this.
Either don't call braced_list_to_string when processing_template_decl, defer
that only until instantiation time, like:
--- gcc/cp/typeck2.c.jj 2018-09-25 15:14:43.961258143 +0200
+++ gcc/cp/typeck2.c 2018-11-15 17:38:17.471479857 +0100
@@ -807,7 +807,8 @@ store_init_value (tree decl, tree init,
/* Digest the specified initializer into an expression. */
value = digest_init_flags (type, init, flags, tf_warning_or_error);
- if (TREE_CODE (type) == ARRAY_TYPE
+ if (!processing_template_decl
+ && TREE_CODE (type) == ARRAY_TYPE
&& TYPE_STRING_FLAG (TREE_TYPE (type))
&& TREE_CODE (value) == CONSTRUCTOR)
value = braced_list_to_string (type, value);
The question is if there is a guarantee that store_init_value will be called
again during instantiation.
Or revert the:
@@ -1058,9 +1063,7 @@
if (TYPE_PRECISION (typ1) == BITS_PER_UNIT)
{
- if (char_type != char_type_node
- && char_type != signed_char_type_node
- && char_type != unsigned_char_type_node)
+ if (char_type != char_type_node)
{
if (complain & tf_error)
error_at (loc, "char-array initialized from wide string");
hunk of the commit, I don't really see rationale for that change. If as in
this testcase the STRING_CST has unsigned_char_type_node as element type, it
still isn't a wide string. One can also wonder if that
if (char_type == char_type_node)
a few lines later shouldn't be || char_type == signed_char_type_node ||
char_type == unsigned_char_type_node.
