[Bug c++/13558] New: ICE (illegal code) at cp/friend.c:200

[Bug c++/13558] New: ICE (illegal code) at cp/friend.c:200

[pommier at gmm dot insa-tlse dot fr]

the following code snippet:

  class bar { friend class int; };

crashes g++ (GCC) 3.4.0 20031231:
foo.C:1: error: expected identifier
foo.C:1: internal compiler error: tree check: expected class 't', have 'd'
(type_decl) in make_friend_class, at cp/friend.c:200

I don't know if it is related to #3761 or not, so I submit it as a new bug report.

-- 
           Summary: ICE (illegal code) at cp/friend.c:200
           Product: gcc
           Version: 3.4.0
            Status: UNCONFIRMED
          Severity: minor
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pommier at gmm dot insa-tlse dot fr
                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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-03 21:36 -------
>From Phil's regression hunter: Search converges between 2002-12-27-trunk (#177) and 2002-12
-28-trunk (#178).

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |ice-on-valid-code
   Last reconfirmed|0000-00-00 00:00:00         |2004-01-03 21:36:51
               date|                            |
            Summary|ICE (illegal code) at       |[3.4 Regression] ICE
                   |cp/friend.c:200             |(illegal code) at
                   |                            |cp/friend.c:200
   Target Milestone|---                         |3.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-03 22:11 -------
The new parser came in on that date.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-04 04:36 -------
This is not error recovery because the error message is even a regression, I think I have 
a fix for this.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-04 04:37 -------
Created an attachment (id=5406)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=5406&action=view)
Patch which fixes this


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |pinskia at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-04 07:13 -------
Note also this testcase ICE in the same place and the patch fixes it also:
  class bar { friend int; };

The error messages after for the testcase here:
pr13558.cc:1: error: a class-key must be used when declaring a friend
pr13558.cc:1: error: invalid type `int' declared `friend'
For the one in comment #0:
pr13558.cc:1: error: expected identifier
pr13558.cc:1: error: invalid type `int' declared `friend'

I think this is okay error messages and better than what 3.3 gave:
above:
accepted it but warned:
pr13558.cc:1: warning: friend declaration requires class-key, i.e. `friend int'

one in comment #0:
pr13558.cc:1: error: expected identifier
pr13558.cc:1: error: invalid type `int' declared `friend'

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-04 08:02 -------
one more thing this does not ICE with checking disabled.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-checking


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-04 15:01 -------
Patch here: <http://gcc.gnu.org/ml/gcc-patches/2004-01/msg00160.html>.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[gdr at integrable-solutions dot net]

------- Additional Comments From gdr at integrable-solutions dot net  2004-01-04 16:54 -------
Subject: Re:  [3.4 Regression] ICE (illegal code) at cp/friend.c:200

"pinskia at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| Note also this testcase ICE in the same place and the patch fixes it also:
|   class bar { friend int; };
| 
| The error messages after for the testcase here:
| pr13558.cc:1: error: a class-key must be used when declaring a friend

I don't like this new diagnostic because it is misleading.  
We've got a DR resolution that allows things like

   friend typename T::X;

It would be interesting to read the paper in the pre-Kona2003 mailing
by Mike Miller on these issues.

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-05 04:19 -------
I am not taking responsable to change the error message and I am not going to move it to the 
parser also, so I am not working on this any more.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|pinskia at gcc dot gnu dot  |unassigned at gcc dot gnu
                   |org                         |dot org
             Status|ASSIGNED                    |NEW


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-01-09 04:44 -------
Gaby,

which DR are you referring to? There is an old DR (360 IIRC) which says that we 
don't need typename in elaborated type specifiers, thus allowing:

friend class T::X;

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |giovannibajo at libero dot
                   |                            |it, gdr at integrable-
                   |                            |solutions dot net


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[gdr at integrable-solutions dot net]

------- Additional Comments From gdr at integrable-solutions dot net  2004-01-09 14:11 -------
Subject: Re:  [3.4 Regression] ICE (illegal code) at cp/friend.c:200

"giovannibajo at libero dot it" <gcc-bugzilla@gcc.gnu.org> writes:

| Gaby,
| 
| which DR are you referring to? There is an old DR (360 IIRC) which says that we 
| don't need typename in elaborated type specifiers, thus allowing:
| 
| friend class T::X;

I don't have the exact number.  But in the pre-Kona2003 mailing, there
is a paper by Mike Miller on this issue, which was discussed by both
the Evolution Group and the Core Group; Core adopted a slightly
different version of the resolution adopted by Evolution.  But the
point of the resolution is to allow friend typename T::X.
If you allow friend class T::X and X turns out to be a union, you
get another trouble because every class shall be declared with the 
consistent class-key.  The resolution went further to allow

   friend T;

where T is a template parameter.  
(My laptop was stolen yesterday and I don't have reference here me,
but the document must be on the WG21 website)

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[gdr at integrable-solutions dot net]

------- Additional Comments From gdr at integrable-solutions dot net  2004-01-10 02:24 -------
Subject: Re:  [3.4 Regression] ICE (illegal code) at cp/friend.c:200

"giovannibajo at libero dot it" <gcc-bugzilla@gcc.gnu.org> writes:

| Gaby,
| 
| which DR are you referring to? There is an old DR (360 IIRC) which says that we 
| don't need typename in elaborated type specifiers, thus allowing:
| 
| friend class T::X;

see
   
   http://anubis.dkuug.dk/jtc1/sc22/wg21/docs/papers/2003/n1520.pdf

that paper was discussed by both EWG and CWG.  The executive summary
is that 

   friend T;

is allowed -- there are some few restrictions on T, but the class-key
"class" is no longer required.  

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-01-10 10:53 -------
Subject: Re:  [3.4 Regression] ICE (illegal code) at cp/friend.c:200

gdr at integrable-solutions dot net wrote:

>    http://anubis.dkuug.dk/jtc1/sc22/wg21/docs/papers/2003/n1520.pdf

Thanks, it's really interesting.

Giovanni Bajo




-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-01-13 00:44 -------
Subject: Bug 13558

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	steven@gcc.gnu.org	2004-01-13 00:44:51

Modified files:
	gcc/cp         : parser.c ChangeLog 

Log message:
	PR c++/13558
	* parser.c (cp_parser_member_declaration): Any non-type is also
	not a class or a function.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/parser.c.diff?cvsroot=gcc&r1=1.151&r2=1.152
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/ChangeLog.diff?cvsroot=gcc&r1=1.3873&r2=1.3874



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2004-01-13 01:03 -------
Whooo-hoooooo! 
 
Gaby, Giovanni, perhaps your discussion on the first diagnostic deserves its 
own PR? 
 
 

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558

[Bug c++/13558] [3.4 Regression] ICE (illegal code) at cp/friend.c:200

[gdr at integrable-solutions dot net]

------- Additional Comments From gdr at integrable-solutions dot net  2004-01-13 01:22 -------
Subject: Re:  [3.4 Regression] ICE (illegal code) at cp/friend.c:200

"steven at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| Gaby, Giovanni, perhaps your discussion on the first diagnostic deserves its 
| own PR? 

Most probably,  my objection still holds.

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13558