template specializiation

template specializiation

[Thomas Neumann]

Hi,

I have a question concerning template specialization where gcc behaves 
different than other compilers. (I already asked this on 
comp.lang.c++.moderated, but got no response).

Is it possible to specialize just a member variable? (example code 
below). The C++ standard is a little bit vague on this or perhaps I just 
didn't get it. Existing C++ compilers seem to be undecided, gcc and 
Microsoft accept the sample code, while icc, Comeau and Borland reject it.

struct A {
     const A* a;
};

template <const char* c> struct B {
     static const A b;
};

extern const char c[1]="";
extern const char d[1]="";

template<> const A B<c>::b = {&B<d>::b};
template<> const A B<d>::b = {&B<c>::b};


I tend to think it is at least not allowed in this form (two
specialization referencing each other), at least icc gives a somewhat
convincing error message. Does anyone have an idea how to fix it? I want
to use template specialization to plug in information in a template
class, the different specializations need to reference each other. I do
not want to specialize the whole class, as most of it would be redundant.
Also I would like to add new specialization without changing header
files, which _seems_ to work using the approach shown above, but would
not work if I had to specialize the whole class.

Thomas

Re: template specializiation

[Eljay Love-Jensen]

Hi Thomas,

Hmmm, unusual situation (for me).

Using...
http://www.comeaucomputing.com/tryitout/

I tend to trust (EDG's front end) Comeau C++'s error message (strict mode, -tused) on this one.

How to unravel this riddle, I'm not sure.  Sorry.

--Eljay

- - - - -
struct A
{
  const A* a;
};

template <const char* c>
struct B
{
  static const A b;
};

extern const char c[1]="";
extern const char d[1]="";

template<>
const A B<c>::b =
{
  &B<d>::b
};
template<>
const A B<d>::b = // Line 21
{
  &B<c>::b
};
- - - - - 

Comeau C/C++ 4.3.3 (Aug  6 2003 15:13:37) for ONLINE_EVALUATION_BETA1
Copyright 1988-2003 Comeau Computing.  All rights reserved.
MODE:strict errors C++

"ComeauTest.c", line 21: error: explicit specialization of member
          "B<c>::b [with c=d]" must precede its first use
  const A B<d>::b = 
                ^

"ComeauTest.c", line 24: warning: parsing restarts here after previous syntax error
  };
   ^

1 error detected in the compilation of "ComeauTest.c".

Re: template specializiation

[corey taylor]

This information from the ISO 14882 spec might help understand:

14.7.3.6

If a template, a member template or the member of a class template is
explicitly specialized then that specialization
shall be declared before the first use of that specialization that
would cause an implicit instantiation
to take place, in every translation unit in which such a use occurs;
no diagnostic is required. If the program
does not provide a definition for an explicit specialization and
either the specialization is used in a
way that would cause an implicit instantiation to take place or the
member is a virtual member function, the
program is ill-formed, no diagnostic required. An implicit
instantiation is never generated for an explicit
specialization that is declared but not defined. [Example:
template<class T> class Array { /* ... */ };
template<class T> void sort(Array<T>& v) { /* ... */ }
void f(Array<String>& v)
{
sort(v); //use primary template
// sort(Array<T>&), T is String
}
template<> void sort<String>(Array<String>& v); // error: specialization
// after use of primary template
template<> void sort<>(Array<char*>& v); // OK: sort<char*> not yet used
—end example]

corey

On 7/7/05, Eljay Love-Jensen <eljay@adobe.com> wrote:
> Hi Thomas,
> 
> Hmmm, unusual situation (for me).
> 
> Using...
> http://www.comeaucomputing.com/tryitout/
> 
> I tend to trust (EDG's front end) Comeau C++'s error message (strict mode, -tused) on this one.
> 
> How to unravel this riddle, I'm not sure.  Sorry.
> 
> --Eljay
> 
> - - - - -
> struct A
> {
>   const A* a;
> };
> 
> template <const char* c>
> struct B
> {
>   static const A b;
> };
> 
> extern const char c[1]="";
> extern const char d[1]="";
> 
> template<>
> const A B<c>::b =
> {
>   &B<d>::b
> };
> template<>
> const A B<d>::b = // Line 21
> {
>   &B<c>::b
> };
> - - - - -
> 
> Comeau C/C++ 4.3.3 (Aug  6 2003 15:13:37) for ONLINE_EVALUATION_BETA1
> Copyright 1988-2003 Comeau Computing.  All rights reserved.
> MODE:strict errors C++
> 
> "ComeauTest.c", line 21: error: explicit specialization of member
>           "B<c>::b [with c=d]" must precede its first use
>   const A B<d>::b =
>                 ^
> 
> "ComeauTest.c", line 24: warning: parsing restarts here after previous syntax error
>   };
>    ^
> 
> 1 error detected in the compilation of "ComeauTest.c".
> 
> 
>

Re: template specializiation

[Thomas Neumann]

Hi,


> If a template, a member template or the member of a class template is
> explicitly specialized then that specialization
> shall be declared before the first use of that specialization that
but how would I do this? For functions this is easy, but I don't think I 
can declare a static member variable without defining it at the same 
time. I tried some combinations (using extern etc.), but always got 
compiler errors.

Thomas

Re: template specializiation

[corey taylor]

Thomas,

Would this not work for you?

struct A
{
 const A* a;
};

template <const char* c>
struct B
{
 static const A b;
};

extern const char c[1]="";
extern const char d[1]="";

template<> const A B<d>::b;
template<> const A B<c>::b;

template<>
const A B<c>::b =
{
 &B<d>::b
};

template<>
const A B<d>::b = 
{
 &B<c>::b
};

corey

On 7/7/05, Thomas Neumann <tneumann@users.sourceforge.net> wrote:
> Hi,
> 
> 
> > If a template, a member template or the member of a class template is
> > explicitly specialized then that specialization
> > shall be declared before the first use of that specialization that
> but how would I do this? For functions this is easy, but I don't think I
> can declare a static member variable without defining it at the same
> time. I tried some combinations (using extern etc.), but always got
> compiler errors.
> 
> Thomas
>

Re: template specializiation

[corey taylor]

Perhaps this should be filed as a bug on GCC since this should error
as Comeau does.

corey

On 7/7/05, Thomas Neumann <tneumann@users.sourceforge.net> wrote:
> Hi,
> 
> > Would this not work for you?
> >
> 
> > template<> const A B<d>::b;
> > template<> const A B<c>::b;
> >
> > template<> const A B<c>::b = { &B<d>::b };
> > template<> const A B<d>::b = { &B<c>::b };
> 
> hmm, perhaps it _should_ work :) Gcc and Comeau accept it, but Microsoft
> and Borland complain about multiple definitions of B<d>::b and B<c>::d.
> Perhaps I see compiler bugs here, but I would think that they are right.
> Any pointers to the standard?
> 
> Thomas
>

Re: template specializiation

[corey taylor]

Thomas,

  Sure.  What version of Visual Studio are you using as I don't know
that I've had to use this in my MSVC development yet.

14.7.3.15

An explicit specialization of a static data member of a template is a
definition if the declaration includes an
initializer; otherwise, it is a declaration. [Note: there is no syntax
for the definition of a static data member
of a template that requires default initialization.
template<> X Q<int>::x;
This is a declaration regardless of whether X can be default
initialized (8.5). ]

corey

On 7/7/05, Thomas Neumann <tneumann@users.sourceforge.net> wrote:
> Hi,
> 
> > Would this not work for you?
> >
> 
> > template<> const A B<d>::b;
> > template<> const A B<c>::b;
> >
> > template<> const A B<c>::b = { &B<d>::b };
> > template<> const A B<d>::b = { &B<c>::b };
> 
> hmm, perhaps it _should_ work :) Gcc and Comeau accept it, but Microsoft
> and Borland complain about multiple definitions of B<d>::b and B<c>::d.
> Perhaps I see compiler bugs here, but I would think that they are right.
> Any pointers to the standard?
> 
> Thomas
>

Re: template specializiation

[Thomas Neumann]

corey taylor wrote:
> Perhaps this should be filed as a bug on GCC since this should error
> as Comeau does.
this is PR 22354 now.

Thomas