[Bug c++/17011] New: [3.4/3.5 regression] invalid default parameter diagnosed too late

[Bug c++/17011] New: [3.4/3.5 regression] invalid default parameter diagnosed too late

[reichelt at gcc dot gnu dot org]

On 3.4 branch and mainline the invalid default parameter is only
diagnosed at instantiation time and not at definition time:

======================================
template<typename> struct A {};

template<typename T> struct B
{
    A<T> a;

    void foo(A<T>* = &a);

    void bar() { foo(); }
};

// template struct B<int>;
======================================

On the 3.3 branch we get the error message already at defintion time:

bug.cc:7: error: invalid use of member `B<T>::a'

-- 
           Summary: [3.4/3.5 regression] invalid default parameter diagnosed
                    too late
           Product: gcc
           Version: 3.5.0
            Status: UNCONFIRMED
          Keywords: accepts-invalid, diagnostic
          Severity: enhancement
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: reichelt at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org,mark at codesourcery dot
                    com
 BugsThisDependsOn: 16929


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/3.5 regression] invalid default parameter diagnosed too late

[reichelt at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |3.4.2 3.5.0
      Known to work|                            |3.3.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/3.5 regression] invalid default parameter diagnosed too late

[bangerth at dealii dot org]

------- Additional Comments From bangerth at dealii dot org  2004-08-12 21:12 -------
I had to do some research when discussing a bug within IBM's xlC compiler a week ago, 
and here simply paste a piece from a mail to them that also seems relevant in the 
present context: 
----------------------------------- 
However, here's the catch: let's assume that we may cause an implicit  
instantiation. Then, 14.7.1/1 says: "The implicit instantiation of a class  
template specialization causes the instantiation of the declarations, but  
not of the definitions or default arguments...".  
 
Furthermore, 14.7.2 says: "Unless a call is to a [...] member function of  
an explicitly specialized class template, a default argument for a [...]  
member function of a class template is implicitly instantiated when the  
function is called in a context that requires the value of the default  
argument." 
 
Additionally, 14.7.1/11 indicates to me that default arguments are only  
evaluated at the point where they are actually used. 
----------------------------------- 
 
This all may or may not be imminent here, but it at least points to the right paragraphs 
that state when exactly a default argument shall be evaluated. 
 
W. 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/3.5 regression] invalid default parameter diagnosed too late

[mark at codesourcery dot com]

------- Additional Comments From mark at codesourcery dot com  2004-08-12 22:20 -------
Subject: Re:  [3.4/3.5 regression] invalid default parameter
 diagnosed too late

bangerth at dealii dot org wrote:

>------- Additional Comments From bangerth at dealii dot org  2004-08-12 21:12 -------
>I had to do some research when discussing a bug within IBM's xlC compiler a week ago, 
>and here simply paste a piece from a mail to them that also seems relevant in the 
>present context: 
>----------------------------------- 
>However, here's the catch: let's assume that we may cause an implicit  
>instantiation. Then, 14.7.1/1 says: "The implicit instantiation of a class  
>template specialization causes the instantiation of the declarations, but  
>not of the definitions or default arguments...".  
> 
>Furthermore, 14.7.2 says: "Unless a call is to a [...] member function of  
>an explicitly specialized class template, a default argument for a [...]  
>member function of a class template is implicitly instantiated when the  
>function is called in a context that requires the value of the default  
>argument." 
> 
>Additionally, 14.7.1/11 indicates to me that default arguments are only  
>evaluated at the point where they are actually used. 
>----------------------------------- 
> 
>This all may or may not be imminent here, but it at least points to the right paragraphs 
>that state when exactly a default argument shall be evaluated. 
>  
>
That's a good point.  I forgot about that.  I'm not sure exactly how it 
plays in with the QoI issue in 17011; it might mean that it would be 
invalid for the compiler to issue a diagnostic, or it might nt.  But, 
it's certainly interesting.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/3.5 regression] invalid default parameter diagnosed too late

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-08-12 22:55 -------
Uhm I'm not sure those paragraphs are pertinent. They say that the default 
arguments are *instantiated* when they are actually used. 

But the PR is about a default argument having a totally wrong syntax: "&a" is 
invalid if a is not a static member, and surely it's type is not A<T>* (in the 
best case, it would have to be "A<T>* B<T>::*", that is a pointer-to-member).

I guess this basic kind of syntax and type checking is totally orthogonal to 
the instantiation time. I incline to confirm this bug.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2004-08-12 22:55:33
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/3.5 regression] invalid default parameter diagnosed too late

[bangerth at dealii dot org]

------- Additional Comments From bangerth at dealii dot org  2004-08-12 23:59 -------
No, the type of &a is of course the same as the type of &this->a, i.e. 
a pointer to an object, not a pointer-to-member. For that you would 
have to use &B<T>::a. The invalid part in the code Volker showed is 
the implicit reference to this->, which is explicitly not allowed 
by the standard. 
 
One of the reasons why you may not want to parse default arguments 
at template definition time is that you cannot know whether this code, 
for example, is going to be valid or not: 
------------------ 
template <typename T> struct X { 
    int g (int = T(1)); 
}; 
------------------ 
(Think a template type T for which there is no constructor that takes 
an integer.) Now, you will say that here we have a dependent call, whereas 
in the original case we do not, but that may actually be hard to 
figure out in some cases. I guess that it's cases like these that led 
the committee to suggest that default arguments are only evaluated at the 
time of use. 
 
W. 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/4.0 regression] invalid default parameter diagnosed too late

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|3.4.2 4.0                   |3.4.2 4.0.0 3.3.4
   Target Milestone|---                         |3.4.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/4.0 regression] invalid default parameter diagnosed too late

[pinskia at gcc dot gnu dot org]

-- 
Bug 17011 depends on bug 16929, which changed state.

Bug 16929 Summary: [3.4/4.0 Regression] ICE on taking address of member in default parameter to member function
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16929

           What    |Old Value                   |New Value
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |FIXED

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/4.0 regression] invalid default parameter diagnosed too late

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-11-01 00:45 -------
Postponed until GCC 3.4.4.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4.3                       |3.4.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/4.0 regression] invalid default parameter diagnosed too late

[lerdsuwa at gcc dot gnu dot org]

------- Additional Comments From lerdsuwa at gcc dot gnu dot org  2004-11-28 06:03 -------
Under investigation.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/4.0 regression] invalid default parameter diagnosed too late

[lerdsuwa at gcc dot gnu dot org]

------- Additional Comments From lerdsuwa at gcc dot gnu dot org  2004-12-03 10:24 -------
Patch submitted:
  http://gcc.gnu.org/ml/gcc-patches/2004-12/msg00255.html

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/4.0 regression] invalid default parameter diagnosed too late

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-12-04 06:45 -------
Subject: Bug 17011

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	lerdsuwa@gcc.gnu.org	2004-12-04 06:45:14

Modified files:
	gcc/cp         : ChangeLog pt.c semantics.c 
	gcc/testsuite  : ChangeLog 
	gcc/testsuite/g++.dg/template: error15.C instantiate3.C 

Log message:
	PR c++/17011, c++/17971
	* pt.c (tsubst_copy) <FIELD_DECL case>: Check and diagnose
	invalid field.
	(tsubst_copy_and_build) <COMPONENT_REF case>: Check
	error_mark_node after member substitution.
	* semantics.c (finish_id_expression): Call
	finish_non_static_data_member for non-dependent FIELD_DECL.
	
	* g++.dg/template/error15.C: Adjust expected error.
	* g++.dg/template/instantiate3.C: Likewise.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/ChangeLog.diff?cvsroot=gcc&r1=1.4514&r2=1.4515
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/pt.c.diff?cvsroot=gcc&r1=1.953&r2=1.954
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/semantics.c.diff?cvsroot=gcc&r1=1.450&r2=1.451
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.4713&r2=1.4714
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/error15.C.diff?cvsroot=gcc&r1=1.1&r2=1.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/instantiate3.C.diff?cvsroot=gcc&r1=1.3&r2=1.4



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/4.0 regression] invalid default parameter diagnosed too late

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-12-04 06:55 -------
Subject: Bug 17011

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_4-branch
Changes by:	lerdsuwa@gcc.gnu.org	2004-12-04 06:55:00

Modified files:
	gcc/cp         : ChangeLog pt.c semantics.c 
	gcc/testsuite  : ChangeLog 
	gcc/testsuite/g++.dg/template: error15.C instantiate3.C 

Log message:
	PR c++/17011, c++/17971
	* pt.c (tsubst_copy) <FIELD_DECL case>: Check and diagnose
	invalid field.
	(tsubst_copy_and_build) <COMPONENT_REF case>: Check
	error_mark_node after member substitution.
	* semantics.c (finish_id_expression): Call
	finish_non_static_data_member for dependent FIELD_DECL.
	
	* g++.dg/template/error15.C: Adjust expected error.
	* g++.dg/template/instantiate3.C: Likewise.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.3892.2.181&r2=1.3892.2.182
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/pt.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.816.2.46&r2=1.816.2.47
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/semantics.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.381.4.15&r2=1.381.4.16
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.3389.2.317&r2=1.3389.2.318
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/error15.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.1.2.1&r2=1.1.2.2
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/instantiate3.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.2&r2=1.2.20.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011

[Bug c++/17011] [3.4/4.0 regression] invalid default parameter diagnosed too late

[lerdsuwa at gcc dot gnu dot org]

------- Additional Comments From lerdsuwa at gcc dot gnu dot org  2004-12-04 07:06 -------
Fixed in 3.4 branch and mainline.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17011