Largest value of an unsigned integral type

[Dave]

Hello all,

Suppose you have an unsigned integral type T. It's not one of the built-in
types, but rather typedefed off of one of the built-in unsigned integral
types (but we don't know which one).

I want to find the maximum value for this type. This seems to work just
fine:

static_cast<T>(-1)

Is there any reason this should be avoided? Or is this indeed guaranteed to
do the job on a standard-conforming compiler?

I'm not using numeric_limits< > to get the maximum value because, as stated
above, I don't know the underlying built-in type, so I don't know which
specialization of numeric_limits< > to use.

In case anyone is wondering, the type I'm actually working with is
string::size_ty pe. I need to get the maximum value for this type, but,
according to Josuttis, I can only be assured the underlying type is unsigned
integral, but there are no guarantees as to which of the unsigned integral
types it is...

Does this look clean to everyone?

Thanks,
Dave

[Jack Klein]

Re: Largest value of an unsigned integral type

On Thu, 6 Nov 2003 20:53:30 -0700, "Dave" <better_cs_now@ yahoo.com>
wrote in comp.lang.c++:
[color=blue]
>
> Hello all,
>
> Suppose you have an unsigned integral type T. It's not one of the built-in
> types, but rather typedefed off of one of the built-in unsigned integral
> types (but we don't know which one).
>
> I want to find the maximum value for this type. This seems to work just
> fine:
>
> static_cast<T>(-1)
>
> Is there any reason this should be avoided? Or is this indeed guaranteed to
> do the job on a standard-conforming compiler?[/color]

This is guaranteed to work. The conversion of any a value of any
integer type to any unsigned integer type is well defined. If the
value being assigned is outside the range of the destination unsigned
type, it is adjusted by repeatedly adding or subtracting (UTYPE_MAX +
1) until the value is within the range [0...UTYPE_MAX], so converting
-1 to any unsigned yields the maximum value.

--
Jack Klein
Home: http://JK-Technology.Com
FAQs for
comp.lang.c http://www.eskimo.com/~scs/C-faq/top.html
comp.lang.c++ http://www.parashift.com/c++-faq-lite/
alt.comp.lang.l earn.c-c++ ftp://snurse-l.org/pub/acllc-c++/faq

[Andrey Tarasevich]

Re: Largest value of an unsigned integral type

Dave wrote:[color=blue]
> ...
> Suppose you have an unsigned integral type T. It's not one of the built-in
> types, but rather typedefed off of one of the built-in unsigned integral
> types (but we don't know which one).
>
> I want to find the maximum value for this type. This seems to work just
> fine:
>
> static_cast<T>(-1)
>
> Is there any reason this should be avoided? Or is this indeed guaranteed to
> do the job on a standard-conforming compiler?[/color]

Yes, it is guaranteed to do the job.
[color=blue]
> I'm not using numeric_limits< > to get the maximum value because, as stated
> above, I don't know the underlying built-in type, so I don't know which
> specialization of numeric_limits< > to use.[/color]

I don't understand why do you care about knowing "the underlying
built-in type". From the above 'static_cast' example it follows that you
know the typedef-name of the type, don't you? You can immediately use it
with 'std::numeric_l imits'

std::numeric_li mits<T>::max();
[color=blue]
> In case anyone is wondering, the type I'm actually working with is
> string::size_ty pe. I need to get the maximum value for this type, but,
> according to Josuttis, I can only be assured the underlying type is unsigned
> integral, but there are no guarantees as to which of the unsigned integral
> types it is...[/color]

You can simply use 'std::numeric_l imits<std::stri ng::size_type>: :max()'.

Actually, 'std::string::n pos' is guaranteed to represent the largest
value of 'std::string::s ize_type' (it is initialized by using the same
trick with '-1')

--
Best regards,
Andrey Tarasevich

[Jakob Bieling]

Re: Largest value of an unsigned integral type

"Dave" <better_cs_now@ yahoo.com> wrote in message
news:vqm5u9jsln i20d@news.super news.com...[color=blue]
>
> Hello all,
>
> Suppose you have an unsigned integral type T. It's not one of the[/color]
built-in[color=blue]
> types, but rather typedefed off of one of the built-in unsigned integral
> types (but we don't know which one).
>
> I want to find the maximum value for this type. This seems to work just
> fine:
>
> static_cast<T>(-1)
>
> Is there any reason this should be avoided? Or is this indeed guaranteed[/color]
to[color=blue]
> do the job on a standard-conforming compiler?
>
> I'm not using numeric_limits< > to get the maximum value because, as stated
> above, I don't know the underlying built-in type, so I don't know which
> specialization of numeric_limits< > to use.[/color]

Sure you do. numeric_limits <T> works. If this would not work, then the
static_cast would not work either, there, too, you have to know T at compile
time. Think of it as a template function. Maybe it is clearer then.
[color=blue]
> In case anyone is wondering, the type I'm actually working with is
> string::size_ty pe. I need to get the maximum value for this type, but,
> according to Josuttis, I can only be assured the underlying type is[/color]
unsigned[color=blue]
> integral, but there are no guarantees as to which of the unsigned integral
> types it is...
>
> Does this look clean to everyone?[/color]

hth
--
jb

(replace y with x if you want to reply by e-mail)