Re: [C++ PATCH,committed] Fix PR12924 (TEMPLATE_ID_EXPR handlingin finish_class_member_access_expr)

[Kriang Lerdsuwanakij <lerdsuwa at users dot sf dot net>]

Gabriel Dos Reis wrote:
> Kriang Lerdsuwanakij <lerdsuwa@users.sourceforge.net> writes:
>
> | Hi
> | 
> | This obvious patch fix the ICE PR12924 which is a regression in GCC 3.4.
> | We simply forget to deal with TEMPLATE_ID_EXPR that has OVERLOAD,
> | TEMPLATE_DECL, as its first operand besides IDENTIFIER_NODE.
>
> I think the documentation says somewhere that the first operand is an
> IDENTIFIER_NODE.
>
> -- Gaby

I don't think so.  Here is the piece of comment from cp-tree.def,
which should be the most up-to-date compare to other documentations:

/* A template-id, like foo<int>.  The first operand is the template.
   The second is NULL if there are no explicit arguments, or a
   TREE_VEC of arguments.  The template will be a FUNCTION_DECL,
   TEMPLATE_DECL, or an OVERLOAD.  If the template-id refers to a
   member template, the template may be an IDENTIFIER_NODE.  */

Note that there is still some problem deciding whether the scope of
a template-id is template parameter dependent (like PR11814) in GCC now.
Ideally, the first operand is IDENTIFIER_NODE only when the scope is
dependent and *_DECL or OVERLOAD otherwise.

--Kriang