[Bug c++/11752] New: partial ordering overloaded function templates

[Bug c++/11752] New: partial ordering overloaded function templates

[nbecker at fred dot net]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752

           Summary: partial ordering overloaded function templates
           Product: gcc
           Version: 3.2.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: nbecker at fred dot net
                CC: gcc-bugs at gcc dot gnu dot org

In the following example, I believe the call should not be ambiguous, because 
the second overloaded Rnd function is more specialized, and partial ordering 
should apply: 
 
template<int shift, int size, typename T> 
inline T Rnd (T x) { 
  T y = x >> (shift - 1); 
  if (y == Sint<size+1>::max) 
    return y >> 1; 
  else 
    return (y + 1) >> 1; 
} 
 
template<int shift, int size, typename FLT> 
inline std::complex<FLT> Rnd (std::complex<FLT> x) { 
  return std::complex<FLT> (Rnd<size, shift, FLT> (real (x)), Rnd<size, shift, 
FLT> (imag (x))); 
} 
 
../Test/Test6.cc:224: call of overloaded `Rnd(std::complex<int>)' is ambiguous 
../src/fixed/FixedUtil.H:18: candidates are: T Rnd(T) [with int shift = 6, int  
   size = 8, T = std::complex<int>] 
../src/fixed/FixedUtil.H:27:                 std::complex<FLT>  
   Rnd(std::complex<FLT>) [with int shift = 6, int size = 8, FLT = int] 
gcc -v 
Reading specs from /usr/lib/gcc-lib/i386-redhat-linux/3.2.2/specs 
Configured with: ../configure --prefix=/usr --mandir=/usr/share/man 
--infodir=/usr/share/info --enable-shared --enable-threads=posix 
--disable-checking --with-system-zlib --enable-__cxa_atexit 
--host=i386-redhat-linux 
Thread model: posix 
gcc version 3.2.2 20030222 (Red Hat Linux 3.2.2-5)

[Bug c++/11752] partial ordering overloaded function templates

[pinskia at physics dot uc dot edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752


pinskia at physics dot uc dot edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
   Target Milestone|3.4                         |---


------- Additional Comments From pinskia at physics dot uc dot edu  2003-07-31 19:08 -------
Can you provide the preprocessed source file?
Read <http://gcc.gnu.org/bugs.html> for how to produce that file.

[Bug c++/11752] partial ordering overloaded function templates

[nbecker at hns dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752



------- Additional Comments From nbecker at hns dot com  2003-07-31 19:16 -------
Subject: Re:  partial ordering overloaded function templates

On Thursday 31 July 2003 03:08 pm, you wrote:
> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752
>
>
> pinskia at physics dot uc dot edu changed:
>
>            What    |Removed                     |Added
> ---------------------------------------------------------------------------
>- Status|UNCONFIRMED                 |WAITING
>    Target Milestone|3.4                         |---
>
>
> ------- Additional Comments From pinskia at physics dot uc dot edu 
> 2003-07-31 19:08 ------- Can you provide the preprocessed source file?
> Read <http://gcc.gnu.org/bugs.html> for how to produce that file.
>

OK, done.

[Bug c++/11752] partial ordering overloaded function templates

[nbecker at fred dot net]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752


nbecker at fred dot net changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW

[Bug c++/11752] partial ordering overloaded function templates

[pinskia at physics dot uc dot edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752


pinskia at physics dot uc dot edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
         Resolution|INVALID                     |


------- Additional Comments From pinskia at physics dot uc dot edu  2003-08-04 14:58 -------
unconfirmed.

[Bug c++/11752] partial ordering overloaded function templates

[pinskia at physics dot uc dot edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752


pinskia at physics dot uc dot edu changed:

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


------- Additional Comments From pinskia at physics dot uc dot edu  2003-08-04 14:58 -------
Changing to invalid to change to ...

[Bug c++/11752] partial ordering overloaded function templates

[pinskia at physics dot uc dot edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752



------- Additional Comments From pinskia at physics dot uc dot edu  2003-08-04 19:44 -------
ICC 6.0 in strict mode rejects this:

pr11752.cc(29): error: more than one instance of overloaded function "Rnd" matches the 
argument list:
            function template "Rnd<shift,size,T>(T)"
            function template "Rnd<shift,size,FLT>(std::complex<FLT>)"
            argument types are: (ComplexI)
    ComplexI z = Rnd <6, 4> (x);
                 ^

compilation aborted for pr11752.cc (code 2)

[Bug c++/11752] partial ordering overloaded function templates

[lerdsuwa at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752


lerdsuwa at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID


------- Additional Comments From lerdsuwa at gcc dot gnu dot org  2003-08-05 16:29 -------
Not a bug.  According to the standard, partial ordering is defined
based on template argument deduction.  Since the parameter 'size'
is not used anywhere in function parameter or return type, it's
not deducable and partial ordering fails.  So both are treated as
equal, none is more specialized than the other.

[Bug c++/11752] partial ordering overloaded function templates

[bangerth at dealii dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752



------- Additional Comments From bangerth at dealii dot org  2003-08-05 16:32 -------
Here's a self-contained testcase:
-----------------------------
template <typename> struct X {};

template<int shift, typename T>  T    f (T);
template<int shift, typename T>  X<T> f (X<T>);

int main () {
  X<int> x;
  f<6> (x);
}
------------------------------

It fails with 3.4 and icc7:
g/x> /home/bangerth/bin/gcc-3.4-pre/bin/c++ -c bug.cc
bug.cc: In function `int main()':
bug.cc:8: error: call of overloaded `f(X<int>&)' is ambiguous
bug.cc:3: note: candidates are: T f(T) [with int shift = 6, T = X<int>]
bug.cc:4: note:                 X<T> f(X<T>) [with int shift = 6, T = int]
g/x>
g/x> icc -c bug.cc
bug.cc(8): error: more than one instance of overloaded function "f" matches the
argument list:
            function template "f<shift,T>(T)"
            function template "f<shift,T>(X<T>)"
            argument types are: (X<int>)
    f<6> (x);
    ^

compilation aborted for bug.cc (code 2)

I recall that we had another PR like this, where the existence of the
first template arg made a difference, and I remember I didn't understand
the reasoning why that was deemed illegal. Note that if the int template
arg is removed (along with the <6> on the call of f), then both
compilers grok the code.

Someone with more knowledge will have to look at this.

W.

[Bug c++/11752] partial ordering overloaded function templates

[nbecker at hns dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752



------- Additional Comments From nbecker at hns dot com  2003-08-05 16:50 -------
Subject: Re:  partial ordering overloaded function templates

On Tuesday 05 August 2003 12:29 pm, you wrote:
> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11752
>
>
> lerdsuwa at gcc dot gnu dot org changed:
>
>            What    |Removed                     |Added
> ---------------------------------------------------------------------------
>- Status|UNCONFIRMED                 |RESOLVED
>          Resolution|                            |INVALID
>
>
> ------- Additional Comments From lerdsuwa at gcc dot gnu dot org 
> 2003-08-05 16:29 ------- Not a bug.  According to the standard, partial
> ordering is defined based on template argument deduction.  Since the
> parameter 'size'
> is not used anywhere in function parameter or return type, it's
> not deducable and partial ordering fails.  So both are treated as
> equal, none is more specialized than the other.
>
>
>

Thank you for clarifying this.  Unfortunately, I don't own a copy of the 
standard.