[boost] Re: lexical_cast<>() and character traits

From: Gennaro Prota (gennaro_prota_at_[hidden])
Date: 2002-07-24 16:31:56

On Wed, 24 Jul 2002 03:43:31 +0200, Terje Slettebø
<tslettebo_at_[hidden]> wrote:

>>From: "Gennaro Prota" <gennaro_prota_at_[hidden]>
>
>By the way, there's a question to Gennadiy in this posting, as well.
>

Well, if the question is the same for both, the syntax is:

ask /to genna* <your_question>       ;-)

Seriously, I've looked at the new version and I find it very nice.
Only two points:

1) About pointer_to_char_base: maybe someone will argue whether
something like

template<class Target, class Source>
    struct pointer_to_char_base
    {
      static Target do_cast(Source arg)
      {
        return arg[0];
      }
    };

isn't preferable.

In any case, the current implementation triggers undefined behavior
with empty source strings; a trivial fix is:

    template<class Target, class Source>
    struct pointer_to_char_base
    {
      static Target do_cast(Source arg)
      {
        Target t = arg[0];
        if(t != 0  &&  arg[1] != 0)
          throw bad_lexical_cast();

        return t;
      }
    };

2) Stupid question of the day: is there any reason why all Source
function parameters cannot be declared as Source const &?

P.S.: Does anyone know if Intel C++ 6.0 for Windows can properly
support wchar_t as a distinct type? I've also experimented with the
new /Zc:wchar_t switch. Well, if you try the following:

#include <iostream>

void dummy (unsigned short par) {

    std::cout << "wrong overload... - ";
    std::cout << par;

}

void dummy (wchar_t par) {
    std::wcout << L"wchar_t! - ";
    std::wcout << par;
}

int main() {
  dummy (L'1');
}

it (correctly) compiles only with the switch, but (incorrectly) prints
"wrong overload... - 49"! :-(

Genny.