Patchwork C++ PATCH for c++/50852, 53039 (problems with typedefs in templates)

login
register
mail settings
Submitter Jason Merrill
Date July 5, 2012, 9:24 p.m.
Message ID <4FF605F9.1040102@redhat.com>
Download mbox | patch
Permalink /patch/169274/
State New
Headers show

Comments

Jason Merrill - July 5, 2012, 9:24 p.m.
On 07/05/2012 03:37 PM, Jason Merrill wrote:
> have not done that for non-type arguments so far.  This patch implements
> typedef stripping for non-type template arguments, and thereby avoids
> the need for any fixup of sibling information.
>
> The patch also fixes 53039, which is another case of treating things as
> interchangeable that turn out not to be: in this case they are template
> parameter packs that are compared for identity in arg_from_parm_pack_p,
> which was running into problems during template parameter fixup.  I
> didn't expect this patch to fix the bug, but apparently removing the
> fixup solves the problem for this testcase.

And since we no longer have to try to deal with comparing fixed-up and 
non-fixed-up template parms, we can revert the earlier patch for PR 
46394 which led to the issue in 53039.

Tested x86_64-pc-linux-gnu, applying to trunk.

Patch

commit 1213de7ecf31a323c64fa75fba30e523431602db
Author: Jason Merrill <jason@redhat.com>
Date:   Thu Jul 5 15:49:23 2012 -0400

    	PR c++/53039
    	* pt.c (arg_from_parm_pack_p): Go back to using same_type_p or
    	cp_tree_equal.

diff --git a/gcc/cp/pt.c b/gcc/cp/pt.c
index e07a362..df5d1f6 100644
--- a/gcc/cp/pt.c
+++ b/gcc/cp/pt.c
@@ -3760,34 +3760,13 @@  arg_from_parm_pack_p (tree arg_pack, tree parm_pack)
     {
       tree expansion = TREE_VEC_ELT (ARGUMENT_PACK_ARGS (arg_pack), 0);
       tree pattern = PACK_EXPANSION_PATTERN (expansion);
-      /* So we have an argument_pack<P...>.  We want to test if P
-	 is actually PARM_PACK.  We will not use cp_tree_equal to
-	 test P and PARM_PACK because during type fixup (by
-	 fixup_template_parm) P can be a pre-fixup version of a
-	 type and PARM_PACK be its post-fixup version.
-	 cp_tree_equal would consider them as different even
-	 though we would want to consider them compatible for our
-	 precise purpose here.
-
-	 Thus we are going to consider that P and PARM_PACK are
-	 compatible if they have the same DECL.  */
-      if ((/* If ARG_PACK is a type parameter pack named by the
-	      same DECL as parm_pack ...  */
-	   (TYPE_P (pattern)
-	    && TYPE_P (parm_pack)
-	    && TYPE_NAME (pattern) == TYPE_NAME (parm_pack))
-	   /* ... or if PARM_PACK is a non-type parameter named by the
-	      same DECL as ARG_PACK.  Note that PARM_PACK being a
-	      non-type parameter means it's either a PARM_DECL or a
-	      TEMPLATE_PARM_INDEX.  */
-	   || (TREE_CODE (pattern) == TEMPLATE_PARM_INDEX
-	       && ((TREE_CODE (parm_pack) == PARM_DECL
-		    && (TEMPLATE_PARM_DECL (pattern)
-			== TEMPLATE_PARM_DECL (DECL_INITIAL (parm_pack))))
-		   || (TREE_CODE (parm_pack) == TEMPLATE_PARM_INDEX
-		       && (TEMPLATE_PARM_DECL (pattern)
-			   == TEMPLATE_PARM_DECL (parm_pack))))))
-	  && template_parameter_pack_p (pattern))
+      if ((TYPE_P (pattern) && same_type_p (pattern, parm_pack))
+	  || (!TYPE_P (pattern) && cp_tree_equal (parm_pack, pattern)))
+	/* The argument pack that the parameter maps to is just an
+	   expansion of the parameter itself, such as one would
+	   find in the implicit typedef of a class inside the
+	   class itself.  Consider this parameter "unsubstituted",
+	   so that we will maintain the outer pack expansion.  */
 	return true;
     }
   return false;