Lars Gullik Bjønnes wrote:
> Peter Kümmel <[EMAIL PROTECTED]> writes:
>
> | When we use a C compiler then it uses the C-style cast (bool),
> | so I think the correctest way is to use static_cast<bool>,
> | but this is so ugly that I would prefer (bool).
>
> C-style casts are just too powerful... don't use them.
I agree.
> | I'm not sure if the bool constructor -bool::bool(int)-
> | is really as fast as the casts.
>
> I am. If you are not, please provide measurements.
When you are sure, I also would prefer bool().
> | But sometime it would simple help to "code what you mean" :)
> |
> | Here an example:
> |
> | enum type{...}
> | bool match(type a, type b) { return (a & b); }
> |
> | Does this not mean a==b? So why not write it down?
> | So we will get a boolean without any casts.
>
> That depends.... it could very well mean ((a & b) == b)
Yes, I forgot 11 & 10 = 10 > 0 = true.
> (and also if (a & b) is not b, then (a & b) is 0, so the explict
> comparison should not be needed.)
Does the compiler not always generates a comparison with 0?
(Maybe because of this the warning)
> If a is a bit-field the (a & b) is a pretty usual construct.
> (bit-field in the 1,2,4,8,16 sense)
