[Bug c++/11993] New: gcc finds passing reference of template argument as template argument invalid

[Bug c++/11993] New: gcc finds passing reference of template argument as template argument invalid

[stefaandr at hotmail dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993

           Summary: gcc finds passing reference of template argument as
                    template argument invalid
           Product: gcc
           Version: 3.4
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: stefaandr at hotmail dot com
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu

this code doesn't compile anymore on 3.4, it does on 3.3.1 and before (but I
have no idea if this is a valid or invalid decision or 3.4)

Reading specs from /esat/quendi/gcc/lib/gcc/i686-pc-linux-gnu/3.4/specs
Configured with: ../../gcc/3.4/configure --program-suffix=-3.4
--enable-languages=c,c++ --prefix=/esat/quendi/gcc : (reconfigured)
Thread model: posix
gcc version 3.4 20030820 (experimental)
 /esat/quendi/gcc/libexec/gcc/i686-pc-linux-gnu/3.4/cc1plus -E -quiet -v
-D_GNU_SOURCE /tmp/test5.cpp -mtune=pentiumpro -o test5.ii
ignoring nonexistent directory
"/esat/quendi/gcc/lib/gcc/i686-pc-linux-gnu/3.4/../../../../i686-pc-linux-gnu/include"
#include "..." search starts here:
#include <...> search starts here:
 /esat/quendi/gcc/lib/gcc/i686-pc-linux-gnu/3.4/../../../../include/c++/3.4
 /esat/quendi/gcc/lib/gcc/i686-pc-linux-gnu/3.4/../../../../include/c++/3.4/i686-pc-linux-gnu
 /esat/quendi/gcc/lib/gcc/i686-pc-linux-gnu/3.4/../../../../include/c++/3.4/backward
 /usr/local/include
 /esat/quendi/gcc/include
 /esat/quendi/gcc/lib/gcc/i686-pc-linux-gnu/3.4/include
 /usr/include
End of search list.
 /esat/quendi/gcc/libexec/gcc/i686-pc-linux-gnu/3.4/cc1plus -fpreprocessed
test5.ii -quiet -dumpbase test5.cpp -mtune=pentiumpro -auxbase-strip
/tmp/test5.o -version -o test5.s
GNU C++ version 3.4 20030820 (experimental) (i686-pc-linux-gnu)
        compiled by GNU C version 3.3.1.
GGC heuristics: --param ggc-min-expand=55 --param ggc-min-heapsize=48056
/tmp/test5.cpp:2: error: template argument 1 is invalid
/tmp/test5.cpp:2: error: expected `{'
/tmp/test5.cpp:2: error: expected init-declarator
/tmp/test5.cpp:2: error: expected `;'

test5.ii:
# 1 "/tmp/test5.cpp"
# 1 "<built-in>"
# 1 "<command line>"
# 1 "/tmp/test5.cpp"
template <int * a> class class2 {};
template <int & a> class class1 : public class2 <&a> {};

[Bug c++/11993] gcc finds passing reference of template argument as template argument invalid

[bangerth at dealii dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993


bangerth at dealii dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2003-08-20 14:36:07
               date|                            |


------- Additional Comments From bangerth at dealii dot org  2003-08-20 14:36 -------
Confirmed. I tend to think that it might somehow be related to PR 11847. 
 
W.

[Bug c++/11993] [3.4 regression] gcc finds passing reference of template argument as template argument invalid

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |rejects-valid


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-08-20 14:42 -------
Even more evidence that this is related to bug 11847, the same date from the Phil's regression 
hunter: Search converges between 2003-07-08-trunk (#337) and 2003-07-09-trunk (#338).

[Bug c++/11993] [3.4 regression] gcc finds passing reference of template argument as template argument invalid

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
           Severity|normal                      |critical
           Priority|P2                          |P1
   Last reconfirmed|2003-08-20 14:36:07         |2003-08-23 03:14:33
               date|                            |

[Bug c++/11993] [3.4 regression] gcc finds passing reference of template argument as template argument invalid

[janis187 at us dot ibm dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993


janis187 at us dot ibm dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mmitchel at gcc dot gnu dot
                   |                            |org


------- Additional Comments From janis187 at us dot ibm dot com  2003-08-29 20:17 -------
The regression in PR 11993 was introduced or exposed by this patch
(not the same one that introduced or exposed PR 11847, which was added
a few hours earlier):

--- gcc/gcc/ChangeLog ---

  2003-07-08  Mark Mitchell  <mark@codesourcery.com>

        * fold-const.c (make_range): Do not access operand 1 for a
        zero-operand operator.

--- gcc/gcc/cp/ChangeLog ---

  2003-07-08  Mark Mitchell  <mark@codesourcery.com>

        * cp-tree.def (NON_DEPENDENT_EXPR): New node.
        * cp-tree.h (build_call_from_tree): Remove.
        (build_member_call): Likewise.
        (dependent_template_arg_p): Remove.

        [lots more]

The regression hunt took place on i686-pc-linux-gnu using the submitter's
test case.

[Bug c++/11993] [3.4 regression] gcc finds passing reference of template argument as template argument invalid

[mmitchel at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993


mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID


------- Additional Comments From mmitchel at gcc dot gnu dot org  2003-08-30 00:34 -------
This code is invalid.  

The standard is very clear about what form a non-type template argument may
take, and address-of-nontype-template-parameter is not one of the allowed cases.  

See [temp.arg.nontype].

Arguably, this is a defect, but there has been no DR filed with the committee
about this issue.

[Bug c++/11993] [3.4 regression] gcc finds passing reference of template argument as template argument invalid

[stefaandr at hotmail dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993



------- Additional Comments From stefaandr at hotmail dot com  2003-08-30 10:17 -------
Subject: Re:  [3.4 regression] gcc finds passing reference of template arg

>The standard is very clear about what form a non-type template argument may
>take, and address-of-nontype-template-parameter is not one of the allowed 
>cases.
The same section you reference to contains:
A template-argument for a non-type, non-template template-parameter shall be 
one of:
...
- the address of an object or function with external linkage (excluding 
non-static class-members...), expressed as &id-expression...

For clarity, the code in question is the inheritance part: "public class2 
<&a> {};"
To me, this seems the case above, because I think in the definition of the 
templated class class1, "a" is "the name of an object or function with 
external linkage..." as mentioned in the same paragraph (and I think it must 
be, because you didn't question the definition of class1, and it must comply 
to the same rules in the paragraph).  The fact that the name of any 
non-template parameter is also allowed, doesn't contradict this in my 
opinion.

_________________________________________________________________
MSN 8: Get 6 months for $9.95/month. http://join.msn.com/?page=dept/dialup

[Bug c++/11993] [3.4 regression] gcc finds passing reference of template argument as template argument invalid

[mark at codesourcery dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993



------- Additional Comments From mark at codesourcery dot com  2003-08-30 19:17 -------
Subject: Re:  [3.4 regression] gcc finds passing reference of
	template argument as template argument invalid

On Sat, 2003-08-30 at 03:17, stefaandr at hotmail dot com wrote:
> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993
> 
> 
> 
> ------- Additional Comments From stefaandr at hotmail dot com  2003-08-30 10:17 -------
> Subject: Re:  [3.4 regression] gcc finds passing reference of template arg
> 
> >The standard is very clear about what form a non-type template argument may
> >take, and address-of-nontype-template-parameter is not one of the allowed 
> >cases.
> The same section you reference to contains:
> A template-argument for a non-type, non-template template-parameter shall be 
> one of:
> ...
> - the address of an object or function with external linkage (excluding 
> non-static class-members...), expressed as &id-expression...
> 
> For clarity, the code in question is the inheritance part: "public class2 
> <&a> {};"
> To me, this seems the case above, because I think in the definition of the 
> templated class class1, "a" is "the name of an object or function with 
> external linkage..." as mentioned in the same paragraph (and I think it must 
> be, because you didn't question the definition of class1, and it must comply 
> to the same rules in the paragraph).  The fact that the name of any 
> non-template parameter is also allowed, doesn't contradict this in my 
> opinion.

Actually, "a" is the name of a nontype template parameter.

Since the type of that nontype template parameter is a pointer type,
that nontype template parameter must in fact be bound to an object or
function with external linkage.  But, the nontype template parameter is
not itself such an object.

That's why I said that this should perhaps be considered a defect in the
standard.  I'd suggest raising this on comp.std.c++; if the consensus is
that this is a defect, someone will open a DR.  G++ generally implements
corrections for DRs.

Yours,

[Bug c++/11993] [3.4 regression] gcc finds passing reference of template argument as template argument invalid

[stefaandr at hotmail dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11993



------- Additional Comments From stefaandr at hotmail dot com  2003-09-11 22:36 -------
As suggested, I have posted a message on comp.std.c++ to address this
possibility of a defect in the standard.  

Newsgroups: comp.std.c++
Subject: Passing the address of a non-type template parameter
Date: Mon, 8 Sep 2003 08:28:42 +0000 (UTC)
Message-ID: <1062669363.819321@seven.kulnet.kuleuven.ac.be>
X-Trace: mail2news.demon.co.uk 1063009722 11190 10.0.0.1 (8 Sep 2003 08:28:42 GMT)