[PATCH] PR c++/55311 - Cannot specialize alias template with arg of type array of char

[PATCH] PR c++/55311 - Cannot specialize alias template with arg of type array of char

[Dodji Seketeli]

Hello,

Consider this test case:

     1	template <const char *const C, typename T>
     2	struct A
     3	{};
     4
     5	struct B {};
     6
     7	extern constexpr char HELLO_WORLD[] = "hello world";
     8
     9	A<HELLO_WORLD, B> g; // <-- This works fine
    10
    11	template <typename T>
    12	using PartiallySpecialized = A<HELLO_WORLD, T>;  // <-- This fails
    13

At line 12 G++ fails to instantiate the alias template that has a
string variable initialized with a string literal, with the error
message:

    test.cc:12:46: error: ‘"hello world"’ is not a valid template argument of type ‘const char*’ because ‘"hello world"’ is not a variable
     using PartiallySpecialized = A<HELLO_WORLD, T>;  // <-- This fails
                                              ^

Note that instantiating the template A at line 9 with the same
arguments as in the problematic case above works.

This happens in the context of lookup_template_class_1, when it handles
the alias template instantiation A<HELLO_WORLD, T> and thus passes the
VAR_DECL for HELLO_WORLD to convert_nontype_argument.

Note that from there decay_conversion replaces the the VAR_DECL with
its STRING_CST initializer[1].  Latter on, convert_nontype_argument
checks that the HELLO_WORLD constant it received as argument was
indeed a VAR_DECL:

      else
	{
	  tree decl;

	  decl = ((TREE_CODE (expr) == ADDR_EXPR)
		  ? TREE_OPERAND (expr, 0) : expr);
	  if (TREE_CODE (decl) != VAR_DECL)
	    {
	      error ("%qE is not a valid template argument of type %qT "
		     "because %qE is not a variable",
		     expr, type, decl);
	      return NULL_TREE;
	    }

But the issue is, that VAR_DECL has been replaced by STRING_CST, so
the last 'if' above fails.

The idea of the patch I am proposing is to do the test above on the
argument received by convert_nontype_argument rather than on its
modified (decayed?) form.

[1]: The relacement of the VAR_DECL by its initializer is done by
decay_conversion by callig decl_constant_value_safe.  That replacement
doesn't happen if processing_template_decl is not set.  That's why it
doesn't happen for the class template instantiation at line 9, leading
to no error message there.

Bootstrapped and tested on x86_64-unknown-linux-gnu against trunk.

gcc/cp/

	PR c++/55311
	* pt.c (convert_nontype_argument<array arguments>): Perform the
	sanity tests on the initial expression otherwise decay_conversion
	might have made us loose some information.

gcc/testsuite/

	PR c++/55311
	* g++.dg/cpp0x/alias-decl-30.C: New test.
---
 gcc/cp/pt.c                                |  8 ++++----
 gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C | 15 +++++++++++++++
 2 files changed, 19 insertions(+), 4 deletions(-)
 create mode 100644 gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C

diff --git a/gcc/cp/pt.c b/gcc/cp/pt.c
index 1b3f039..4af30cf 100644
--- a/gcc/cp/pt.c
+++ b/gcc/cp/pt.c
@@ -5454,7 +5454,7 @@ unify_overload_resolution_failure (bool explain_p, tree arg)
 static tree
 convert_nontype_argument (tree type, tree expr, tsubst_flags_t complain)
 {
-  tree expr_type;
+  tree expr_type, initial_expr = expr;
 
   /* Detect immediately string literals as invalid non-type argument.
      This special-case is not needed for correctness (we would easily
@@ -5658,10 +5658,10 @@ convert_nontype_argument (tree type, tree expr, tsubst_flags_t complain)
 	}
       else
 	{
-	  tree decl;
+	  tree decl = STRIP_NOPS (initial_expr);
 
-	  decl = ((TREE_CODE (expr) == ADDR_EXPR)
-		  ? TREE_OPERAND (expr, 0) : expr);
+	  decl = ((TREE_CODE (initial_expr) == ADDR_EXPR)
+		  ? TREE_OPERAND (initial_expr, 0) : initial_expr);
 	  if (TREE_CODE (decl) != VAR_DECL)
 	    {
 	      error ("%qE is not a valid template argument of type %qT "
diff --git a/gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C b/gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C
new file mode 100644
index 0000000..7ad5e6d
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C
@@ -0,0 +1,15 @@
+// Origin PR c++/55311
+// { dg-do compile { target c++11 } }
+
+template <const char *const C, typename T>
+struct A
+{};
+
+struct B {};
+
+extern constexpr char HELLO_WORLD[] = "hello world";
+
+A<HELLO_WORLD, B> g; // <-- This works fine
+
+template <typename T>
+using PartiallySpecialized = A<HELLO_WORLD, T>;  // <-- This fails
-- 
		Dodji

Re: [PATCH] PR c++/55311 - Cannot specialize alias template with arg of type array of char

[Jason Merrill]

On 12/22/2012 11:03 AM, Dodji Seketeli wrote:
> [1]: The relacement of the VAR_DECL by its initializer is done by
> decay_conversion by callig decl_constant_value_safe.  That replacement
> doesn't happen if processing_template_decl is not set.  That's why it
> doesn't happen for the class template instantiation at line 9, leading
> to no error message there.

This is the bug.  decay_conversion on an array should take the address, 
not return the initializer.

Jason

Re: [PATCH] PR c++/55311 - Cannot specialize alias template with arg of type array of char

[Dodji Seketeli]

Jason Merrill <jason@redhat.com> writes:

> On 12/22/2012 11:03 AM, Dodji Seketeli wrote:
>> [1]: The relacement of the VAR_DECL by its initializer is done by
>> decay_conversion by callig decl_constant_value_safe.  That replacement
>> doesn't happen if processing_template_decl is not set.  That's why it
>> doesn't happen for the class template instantiation at line 9, leading
>> to no error message there.
>
> This is the bug.  decay_conversion on an array should take the
> address, not return the initializer.

Right.  So would something like this be acceptable then?

gcc/cp/

	PR c++/55311
	* pt.c (decay_conversion): Do not return the initializer of an array.

gcc/testsuite/

	PR c++/55311
	* g++.dg/cpp0x/alias-decl-30.C: New test.
---
 gcc/cp/typeck.c                            | 20 +++++++++++++-------
 gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C | 15 +++++++++++++++
 2 files changed, 28 insertions(+), 7 deletions(-)
 create mode 100644 gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C

diff --git a/gcc/cp/typeck.c b/gcc/cp/typeck.c
index ef76dcd..3fa913d 100644
--- a/gcc/cp/typeck.c
+++ b/gcc/cp/typeck.c
@@ -1880,19 +1880,25 @@ decay_conversion (tree exp, tsubst_flags_t complain)
       return error_mark_node;
     }
 
-  /* FIXME remove? at least need to remember that this isn't really a
-     constant expression if EXP isn't decl_constant_var_p, like with
-     C_MAYBE_CONST_EXPR.  */
-  exp = decl_constant_value_safe (exp);
-  if (error_operand_p (exp))
-    return error_mark_node;
+  code = TREE_CODE (type);
+
+  /* For and array decl decay_conversion should not try to return its
+     initializer.  */
+  if (code != ARRAY_TYPE)
+    {
+      /* FIXME remove? at least need to remember that this isn't really a
+	 constant expression if EXP isn't decl_constant_var_p, like with
+	 C_MAYBE_CONST_EXPR.  */
+      exp = decl_constant_value_safe (exp);
+      if (error_operand_p (exp))
+	return error_mark_node;
+    }
 
   if (NULLPTR_TYPE_P (type) && !TREE_SIDE_EFFECTS (exp))
     return nullptr_node;
 
   /* build_c_cast puts on a NOP_EXPR to make the result not an lvalue.
      Leave such NOP_EXPRs, since RHS is being used in non-lvalue context.  */
-  code = TREE_CODE (type);
   if (code == VOID_TYPE)
     {
       if (complain & tf_error)
diff --git a/gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C b/gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C
new file mode 100644
index 0000000..7ad5e6d
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp0x/alias-decl-30.C
@@ -0,0 +1,15 @@
+// Origin PR c++/55311
+// { dg-do compile { target c++11 } }
+
+template <const char *const C, typename T>
+struct A
+{};
+
+struct B {};
+
+extern constexpr char HELLO_WORLD[] = "hello world";
+
+A<HELLO_WORLD, B> g; // <-- This works fine
+
+template <typename T>
+using PartiallySpecialized = A<HELLO_WORLD, T>;  // <-- This fails
-- 
		Dodji

Re: [PATCH] PR c++/55311 - Cannot specialize alias template with arg of type array of char

[Jason Merrill]

On 12/24/2012 09:23 AM, Dodji Seketeli wrote:
> +  /* For and array decl decay_conversion should not try to return its

"an array..."

OK with that change.

Jason