Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)

Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)

[jason]

Synopsis: [3.3/3.4 regression] ice in build_baselink (templates)

State-Changed-From-To: analyzed->closed
State-Changed-By: jason
State-Changed-When: Tue Mar 11 22:51:48 2003
State-Changed-Why:
    fixed

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9820

Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)

[jason]

Synopsis: [3.3/3.4 regression] ice in build_baselink (templates)

Responsible-Changed-From-To: unassigned->jason
Responsible-Changed-By: jason
Responsible-Changed-When: Tue Mar 11 20:16:03 2003
Responsible-Changed-Why:
    got it

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9820

Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)

[Janis Johnson]

The following reply was made to PR c++/9820; it has been noted by GNATS.

From: Janis Johnson <janis187@us.ibm.com>
To: gcc-gnats@gcc.gnu.org, gcc-bugs@gcc.gnu.org, nobody@gcc.gnu.org,
   gcc-prs@gcc.gnu.org, poschmid@lbl.gov
Cc:  
Subject: Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)
Date: Tue, 25 Feb 2003 12:43:20 -0800

 Between 2002-10-30 and 2002-12-24, the mainline and the 3.3
 branch (after its creation) compiled the test case with no
 error message, and then the error message came back, along
 with the ICE.
 
 The patch that caused the test case to compile successfully:
 
 2002-10-30  Kriang Lerdsuwanakij  <lerdsuwa@users.sourceforge.net>
 
         Core issue 287, PR c++/7639
         * cp-tree.h (lang_type_class): Add decl_list field.
         (CLASSTYPE_DECL_LIST): New macro.
         (maybe_add_class_template_decl_list): Add declaration.
         * class.c (duplicate_tag_error): Initialize CLASSTYPE_DECL_LIST.
 
         (unreverse_member_declarations): Reverse CLASSTYPE_DECL_LIST.
         (maybe_add_class_template_decl_list): New function.
         (add_implicitly_declared_members): Use it.
         * decl.c (maybe_process_template_type_declaration): Likewise.
         (pushtag): Likewise.
         * friend.c (add_friend): Likewise.
         (make_friend_class): Likewise.
         * semantics.c (finish_member_declaration): Likewise.
         (begin_class_definition): Initialize CLASSTYPE_DECL_LIST.
         * pt.c (instantiate_class_template): Use CLASSTYPE_DECL_LIST
         to process members and friends in the order of declaration.
 
 The patch to mainline and 3.3-branch that brought back the
 error message, plus the ICE:
 
 2002-12-24  Nathan Sidwell  <nathan@codesourcery.com>
 
         PR C++/7964
         * cp-tree.h (resolve_scoped_fn_name): Prototype.
         * call.c (resolve_scoped_fn_name): New function. Deal with
         more template expansion. Broken out of ...
         * parse.y (parse_finish_call_expr): ... here. Call it.
         * decl2.c (build_expr_from_tree, CALL_EXPR): Use
         resolve_scoped_fn_name and build_call_from_tree.
 
 The regression hunt used the original test case, not the
 very short one that Wolfgang provided.
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9820
 
 
 

Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)

[Wolfgang Bangerth]

The following reply was made to PR c++/9820; it has been noted by GNATS.

From: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
To: Janis Johnson <janis187@us.ibm.com>
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)
Date: Mon, 24 Feb 2003 17:27:37 -0600 (CST)

 > > However, Janis, I can't seem to create a testcase that is compilable by 
 > > 3.2 and yet generates the ICE in 3.3 and 3.4. Why don't you just grep the 
 > > output of the compiler?
 > 
 > Yes, that's what I normally do for ice-on-illegal code.  If
 > the submitter's original test case is supposed to compile
 > successfully with earlier versions of GCC, I might end up
 > finding a patch that's not relevant to the submitter's ICE.
 
 Well, it's supposed to, but apparently one bug (not accepting legal 
 syntax) is superceded here by something worse (an ICE)...
 
 W.
 
 -------------------------------------------------------------------------
 Wolfgang Bangerth             email:            bangerth@ticam.utexas.edu
                               www: http://www.ticam.utexas.edu/~bangerth/
 
 

Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)

[Janis Johnson]

The following reply was made to PR c++/9820; it has been noted by GNATS.

From: Janis Johnson <janis187@us.ibm.com>
To: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
Cc: gcc-gnats@gcc.gnu.org, Janis Johnson <janis187@us.ibm.com>
Subject: Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)
Date: Mon, 24 Feb 2003 15:26:40 -0800

 On Mon, Feb 24, 2003 at 05:13:26PM -0600, Wolfgang Bangerth wrote:
 > 
 > Here's a smaller testcase:
 > ------------------------
 > template <typename T> struct X {
 >     template<typename> static int test(...);
 >     template<typename> static int test(int *);
 > 
 >     static const int i = sizeof(X<T>::template test<int>(0));
 > };
 > 
 > const int Yes = X<int>::i;
 > ------------------------
 > 
 > However, Janis, I can't seem to create a testcase that is compilable by 
 > 3.2 and yet generates the ICE in 3.3 and 3.4. Why don't you just grep the 
 > output of the compiler?
 
 Yes, that's what I normally do for ice-on-illegal code.  If
 the submitter's original test case is supposed to compile
 successfully with earlier versions of GCC, I might end up
 finding a patch that's not relevant to the submitter's ICE.

Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)

[Wolfgang Bangerth]

The following reply was made to PR c++/9820; it has been noted by GNATS.

From: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
To: gcc-gnats@gcc.gnu.org, Janis Johnson <janis187@us.ibm.com>
Cc:  
Subject: Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)
Date: Mon, 24 Feb 2003 17:13:26 -0600 (CST)

 Here's a smaller testcase:
 ------------------------
 template <typename T> struct X {
     template<typename> static int test(...);
     template<typename> static int test(int *);
 
     static const int i = sizeof(X<T>::template test<int>(0));
 };
 
 const int Yes = X<int>::i;
 ------------------------
 
 However, Janis, I can't seem to create a testcase that is compilable by 
 3.2 and yet generates the ICE in 3.3 and 3.4. Why don't you just grep the 
 output of the compiler?
 
 W.
 
 -------------------------------------------------------------------------
 Wolfgang Bangerth             email:            bangerth@ticam.utexas.edu
                               www: http://www.ticam.utexas.edu/~bangerth/
 
 

Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)

[Janis Johnson]

The following reply was made to PR c++/9820; it has been noted by GNATS.

From: Janis Johnson <janis187@us.ibm.com>
To: gcc-gnats@gcc.gnu.org, gcc-bugs@gcc.gnu.org, nobody@gcc.gnu.org,
   gcc-prs@gcc.gnu.org, poschmid@lbl.gov
Cc:  
Subject: Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)
Date: Mon, 24 Feb 2003 14:47:31 -0800

 The submitter says this test case should be valid,  but
 GCC 3.0.4 and the 3.2-branch both report errors.  I was
 planning to hunt for the patch that caused the regression,
 but if there's a test case that ICEs on legal code, I'd prefer
 to use that one.
 
 Output from GCC 3.2.2 20021220 (prerelease) with -w:
 
 9820.C: In instantiation of `IsFunctionT<int>':
 9820.C:23:   instantiated from `IsClassT<int>'
 9820.C:23:   instantiated from `void check() [with T = int]'
 9820.C:29:   instantiated from here
 9820.C:23: no method `IsFunctionT<int>::test<int>'
 9820.C:23: enumerator value for `Yes' not integer constant
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9820
 
 
 

Re: c++/9820: [3.3/3.4 regression] ice in build_baselink (templates)

[bangerth]

Old Synopsis: ice in build_baselink (templates)
New Synopsis: [3.3/3.4 regression] ice in build_baselink (templates)

State-Changed-From-To: open->analyzed
State-Changed-By: bangerth
State-Changed-When: Mon Feb 24 15:27:47 2003
State-Changed-Why:
    Confirmed. A regression on 3.3 and mainline w.r.t 3.2

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9820