TITLE: template function partial specialization partial ordering

(Newsgroups: comp.lang.c++.moderated, 8 Apr 2000)

DIMOV: Peter Dimov wrote:

> > template<class T> void f(T);  // #1: template function
> >
> > template<> void f(int*);      // #2: explicit specialization of #1
> >
> > template<class T> void f(T*); // #3: overloads #1
> >
> > int* pi = 0;
> >
> > f(pi); // calls #3, NOT #2! #3 > #1, so the presence of #2 doesn't
> > matter

STEWART: Rob Stewart <donotspamme@giage.com>

> If I understand correctly, only #1 and #3 are compared when
> determine which to instantiate for this call.  If #1 had been
> selected, then #2 could be considered as a better fit of #1 for
> this particular call, right?

DIMOV: "Peter Dimov" <pdimov@mmltd.net>

Right.

Actually the real story is that #1 would have to be instantiated for
T=int* and the compiler would have noticed that the work is already 
done, but the end result is effectively the same: #2 is a better fit 
than #1.


_______________________________________________
cpptips mailing list
http://cpptips.hyperformix.com