Bug 15625 - [3.4/4.0 Regression] friend defined inside a template fails to find static function
Summary: [3.4/4.0 Regression] friend defined inside a template fails to find static fu...
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: c++ (show other bugs)
Version: 3.4.0
: P2 normal
Target Milestone: 3.4.1
Assignee: Mark Mitchell
URL:
Keywords: rejects-valid
Depends on:
Blocks:
 
Reported: 2004-05-24 01:34 UTC by Ivan Godard
Modified: 2004-10-30 21:11 UTC (History)
1 user (show)

See Also:
Host:
Target:
Build:
Known to work: 3.3.3
Known to fail: 3.4.0 4.0.0
Last reconfirmed: 2004-05-24 02:38:53


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Ivan Godard 2004-05-24 01:34:52 UTC
enum Relation {equalOp};
template<typename B>
class A {
public:
    static
    bool    Relop(const A&, const A&, Relation);

    friend
    bool    operator==(const A& a1, const A& a2) {
        return Relop(a1, a2, equalOp);
        }
    B* b;
    };

int main() {
    A<int> a; a == a;
    return 0;
    }


produces:

foo.cc: In function `bool operator==(const A<int>&, const A<int>&)':
foo.cc:16:   instantiated from here
foo.cc:10: error: no matching function for call to `A<B>::Relop(const A<int>&, const A<int>&, Relation)'
foo.cc:6: note: candidates are: static bool A<B>::Relop(const A<B>&, const A<B>&, Relation)


Even if this code is actually invalid (who could tell except the committee?) the error message should at least indicate what is really wrong.

Ivan
Comment 1 Andrew Pinski 2004-05-24 02:38:52 UTC
Confirmed, a regression, this is template related.
Comment 2 Wolfgang Bangerth 2004-05-24 13:57:13 UTC
Here is something slightly smaller: 
------------------- 
template <typename T> struct S { 
    static void foo(const S&); 
    friend void bar(const S& s) { foo(s); } 
}; 
 
int main() { 
    S<int> a; bar(a); 
} 
------------------ 
 
Although all prior versions of gcc before 3.4 and icc in strict mode 
accept this code, I am a little puzzled why this is so: bar() is just 
a friend, not a member, why should it look inside struct S to find the 
static function? Why don't I have to qualify foo as S::foo? 
 
Alas, that isn't the problem, from the error message gcc gives, it 
is pretty obvious that gcc just gets somehow confused about templates 
in this case. 
 
W. 
Comment 3 Mark Mitchell 2004-05-31 18:32:30 UTC
Working on a fix.
Comment 4 Mark Mitchell 2004-05-31 18:34:39 UTC
This code is invalid; according to the standard, the friend function should not
be injected into the namespace scope.  (This is a change from the ARM.) 
However, g++ has always done the injection, and we have not yet removed that
functionality, so this code should work as expected.  

The standard-conforming way of doing this is to have a namespace-scope
declaration of the function.  You can still define it in the class, but it
should be declared outside the class.

This comment is meant to answer Wolfgang's question regarding this seems to work
at all.
Comment 5 GCC Commits 2004-05-31 21:04:24 UTC
Subject: Bug 15625

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_4-branch
Changes by:	mmitchel@gcc.gnu.org	2004-05-31 21:04:15

Modified files:
	gcc/testsuite  : ChangeLog 
	gcc/cp         : ChangeLog call.c cp-tree.h cvt.c name-lookup.c 
	                 pt.c tree.c typeck.c 
	gcc/testsuite/g++.dg/ext: packed3.C packed4.C 
	gcc/testsuite/g++.dg/template: friend.C 
Added files:
	gcc/testsuite/g++.dg/expr: ptrmem5.C 
	gcc/testsuite/g++.dg/ext: packed6.C 
	gcc/testsuite/g++.dg/template: friend27.C friend28.C 
	gcc/testsuite/g++.dg/warn: Wreturn-1.C 

Log message:
	PR c++/15742
	* call.c (build_over_call): Set
	current_function_returns_abnormally even in template functions.
	
	PR c++/15696
	* cp-tree.h (invalid_nonstatic_memfn_p): New function.
	* cvt.c (convert_to_void): Use it.
	* typeck.c (invalid_nonstatic_memfn_p): New function.
	(decay_conversion): Use it.
	
	PR c++/15625
	* pt.c (tsubst_decl): Set DECL_FRIEND_CONTEXT for instantiated
	templates.
	
	PR c++/15629
	* name-lookup.c (arg_assoc_class): Do not find template
	specializations.
	
	PR c++/15209
	* tree.c (lvalue_p_1): Only consider the right-hand side of "."
	expressions when determining whether or not an express is packed.
	
	PR c++/15742
	* g++.dg/warn/Wreturn-1.C: New test.
	
	PR c++/15696
	* g++.dg/expr/ptrmem5.C: New test.
	
	PR c++/15625
	* g++.dg/template/friend27.C: New test.
	
	PR c++/15629
	* g++.dg/template/friend28.C: New test.
	* g++.dg/template/friend.C: Do not depend on <iostream>.  Add
	error message.
	
	PR c++/15209
	* g++.dg/ext/packed3.C: Remove bogus error.
	* g++.dg/ext/packed4.C: Remove bogus check.
	* g++.dg/ext/packed6.C: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.3389.2.187&r2=1.3389.2.188
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.3892.2.109&r2=1.3892.2.110
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/call.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.452.2.16&r2=1.452.2.17
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/cp-tree.h.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.946.4.12&r2=1.946.4.13
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/cvt.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.151.4.1&r2=1.151.4.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/name-lookup.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.34.2.13&r2=1.34.2.14
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/pt.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.816.2.26&r2=1.816.2.27
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/tree.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.360.4.5&r2=1.360.4.6
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/typeck.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.519.2.18&r2=1.519.2.19
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/expr/ptrmem5.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=NONE&r2=1.1.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/ext/packed6.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=NONE&r2=1.1.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/ext/packed3.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.1&r2=1.1.24.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/ext/packed4.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.1&r2=1.1.24.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/friend27.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=NONE&r2=1.1.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/friend28.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=NONE&r2=1.1.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/friend.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.5&r2=1.5.56.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/warn/Wreturn-1.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=NONE&r2=1.1.2.1

Comment 6 GCC Commits 2004-05-31 21:24:37 UTC
Subject: Bug 15625

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	mmitchel@gcc.gnu.org	2004-05-31 21:24:31

Modified files:
	gcc/cp         : call.c cp-tree.h cvt.c name-lookup.c pt.c 
	                 tree.c typeck.c 
	gcc/testsuite/g++.dg/ext: packed3.C packed4.C 
	gcc/testsuite/g++.dg/template: friend.C 
Added files:
	gcc/testsuite/g++.dg/expr: ptrmem5.C 
	gcc/testsuite/g++.dg/ext: packed6.C 
	gcc/testsuite/g++.dg/template: friend27.C friend28.C 
	gcc/testsuite/g++.dg/warn: Wreturn-1.C 

Log message:
	PR c++/15742
	* call.c (build_over_call): Set
	current_function_returns_abnormally even in template functions.
	
	PR c++/15696
	* cp-tree.h (invalid_nonstatic_memfn_p): New function.
	* cvt.c (convert_to_void): Use it.
	* typeck.c (invalid_nonstatic_memfn_p): New function.
	(decay_conversion): Use it.
	
	PR c++/15625
	* pt.c (tsubst_decl): Set DECL_FRIEND_CONTEXT for instantiated
	templates.
	
	PR c++/15629
	* name-lookup.c (arg_assoc_class): Do not find template
	specializations.
	
	PR c++/15209
	* tree.c (lvalue_p_1): Only consider the right-hand side of "."
	expressions when determining whether or not an express is packed.
	
	PR c++/15742
	* g++.dg/warn/Wreturn-1.C: New test.
	
	PR c++/15696
	* g++.dg/expr/ptrmem5.C: New test.
	
	PR c++/15625
	* g++.dg/template/friend27.C: New test.
	
	PR c++/15629
	* g++.dg/template/friend28.C: New test.
	* g++.dg/template/friend.C: Do not depend on <iostream>.  Add
	error message.
	
	PR c++/15209
	* g++.dg/ext/packed3.C: Remove bogus error.
	* g++.dg/ext/packed4.C: Remove bogus check.
	* g++.dg/ext/packed6.C: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/call.c.diff?cvsroot=gcc&r1=1.476&r2=1.477
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/cp-tree.h.diff?cvsroot=gcc&r1=1.969&r2=1.970
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/cvt.c.diff?cvsroot=gcc&r1=1.156&r2=1.157
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/name-lookup.c.diff?cvsroot=gcc&r1=1.56&r2=1.57
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/pt.c.diff?cvsroot=gcc&r1=1.856&r2=1.857
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/tree.c.diff?cvsroot=gcc&r1=1.374&r2=1.375
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/typeck.c.diff?cvsroot=gcc&r1=1.547&r2=1.548
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/expr/ptrmem5.C.diff?cvsroot=gcc&r1=1.1&r2=1.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/ext/packed6.C.diff?cvsroot=gcc&r1=1.1&r2=1.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/ext/packed3.C.diff?cvsroot=gcc&r1=1.1&r2=1.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/ext/packed4.C.diff?cvsroot=gcc&r1=1.1&r2=1.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/friend27.C.diff?cvsroot=gcc&r1=1.1&r2=1.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/friend28.C.diff?cvsroot=gcc&r1=1.1&r2=1.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/friend.C.diff?cvsroot=gcc&r1=1.5&r2=1.6
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/warn/Wreturn-1.C.diff?cvsroot=gcc&r1=1.1&r2=1.2

Comment 7 Mark Mitchell 2004-05-31 21:29:20 UTC
Fixed in GCC 3.4.1.