This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
PR c++/29518
- From: Mark Mitchell <mark at codesourcery dot com>
- To: gcc-patches at gcc dot gnu dot org
- Date: Mon, 13 Nov 2006 09:51:20 -0800
- Subject: PR c++/29518
- Reply-to: mark at codesourcery dot com
This patch fixes (yet another) template static data member problem.
The problem here is that we do not bother to instantiate a static data
member inside sizeof, because we do not need the value of the data
member. However, if we're inside a template parameter list inside a
sizeof, we do need the value. We were already handling this correctly
during parsing, due to:
2005-09-16 Mark Mitchell <mark@codesourcery.com>
PR c++/23914
* parser.c (cp_parser_enclosed_template_argument_list): Make sure
skip_evaluation is false when processing template arguments.
but, we need to do the same thing during substitution into template
parameter lists during template instantiation.
Tested on x86_64-unknown-linux-gnu, applied to mainline. I will apply
to the 4.1 and 4.2 branches after testing completes.
--
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713
2006-11-13 Mark Mitchell <mark@codesourcery.com>
PR c++/29518
* pt.c (coerce_template_parms): Do not skip_evaluation while
substituting template arguments.
2006-11-13 Mark Mitchell <mark@codesourcery.com>
PR c++/29518
* g++.dg/template/static28.C: New test.
Index: gcc/testsuite/g++.dg/template/static28.C
===================================================================
--- gcc/testsuite/g++.dg/template/static28.C (revision 0)
+++ gcc/testsuite/g++.dg/template/static28.C (revision 0)
@@ -0,0 +1,15 @@
+// PR c++/29518
+
+template< bool C > int assertion_failed( int);
+template< class >
+struct N
+{
+ static bool const okay = true;
+ enum {
+ t = sizeof( assertion_failed<okay>( 0))
+ };
+};
+int main()
+{
+ N<int> n;
+}
Index: gcc/cp/pt.c
===================================================================
--- gcc/cp/pt.c (revision 118764)
+++ gcc/cp/pt.c (working copy)
@@ -4131,6 +4131,7 @@ coerce_template_parms (tree parms,
tree inner_args;
tree new_args;
tree new_inner_args;
+ bool saved_skip_evaluation;
inner_args = INNERMOST_TEMPLATE_ARGS (args);
nargs = inner_args ? NUM_TMPL_ARGS (inner_args) : 0;
@@ -4155,6 +4156,10 @@ coerce_template_parms (tree parms,
return error_mark_node;
}
+ /* We need to evaluate the template arguments, even though this
+ template-id may be nested within a "sizeof". */
+ saved_skip_evaluation = skip_evaluation;
+ skip_evaluation = false;
new_inner_args = make_tree_vec (nparms);
new_args = add_outermost_template_args (args, new_inner_args);
for (i = 0; i < nparms; i++)
@@ -4196,6 +4201,7 @@ coerce_template_parms (tree parms,
lost++;
TREE_VEC_ELT (new_inner_args, i) = arg;
}
+ skip_evaluation = saved_skip_evaluation;
if (lost)
return error_mark_node;
@@ -7261,7 +7267,6 @@ tsubst (tree t, tree args, tsubst_flags_
max = tsubst_expr (omax, args, complain, in_decl,
/*integral_constant_expression_p=*/false);
- max = fold_non_dependent_expr (max);
max = fold_decl_constant_value (max);
if (TREE_CODE (max) != INTEGER_CST