Re: Efficient Enum Access...

On 25 Apr., 23:48, Thiago Adams <thiago.ad...@[EMAIL PROTECTED]
> wrote:
> Integers are not directly convertible to enums, for instance:
>
> enum E { e1,  e2 };
>
> E e = 1; // error: cannot convert from 'int' to 'E'
>
> Generally people use static_cast like this:
>
> E e = static_cast<E>(1);
>
> it works, but it is not save.

This is safe in this particular example, but of-course
not in the general case.

> So, my suggestion is to create a similar (but safe) cast called
> "enum_cast"
>
> E e = enum_cast<E>(1);
>
> The implementation would be like this:
>
> //primary template (not implemented)
> template <class T1, class T2> T1 enum_cast(T2);
>
> // and someone must provide the specialization
> template <> E enum_cast(int v)

Why only int as source type?

> {
>    E e = static_cast<E>(v);
>    if (e != e1 || e != e2 )
>      throw std::range_error("invalid convertion from 'int' to 'E'");
>    return e;
>
> };

This definition of enum_cast would break valid
programs, because it's absolutely fine to cast
an integral value into the complete range of the
underlying type of the enumeration as described
by [dcl.enum]/5+6.
An IMO more reasonable definition would just
check whether the given source value is within
the limits

[std::numeric_limits<underlying type of E>::min(),
std::numeric_limits<underlying type of E>::max()]

Unfortunately there exists no general way for
a programmer to determine the exact underlying type
of a classical enumeration type, but it is possible
via a not-too-complicated machinery to find an integral
type that has the same size as it and on the other
hand to find it's integral promotion type (as defined
by [conv.prom]/2).

HTH & Greetings from Bremen,

Daniel Krügler


-- 
      [ See http://www.gotw.ca/resources/clcm.htm
for info about ]
      [ comp.lang.c++.moderated.    First time posters: Do this! ]