«

npos is defined as -1, in both SGI and VS.
So npos+1=0?
I remember that I read a book saying that npos+1==npos.

Alien wrote:


> npos is defined as -1, in both SGI and VS.
> So npos+1=0?
> I remember that I read a book saying that npos+1==npos.

The value of npos is implementation defined. On many platforms it is
-1, but there is no requirement that that be the case. So don't rely
on it.

Best regards,

Tom

Alien wrote:

> npos is defined as -1, in both SGI and VS.

Actually, it is defined that way in the standard [21.3/6]


> So npos+1=0?

As long as npos is an unsigned integral type (which it is): yes.


> I remember that I read a book saying that npos+1==npos.

Well, that would be false.


Best

Kai-Uwe Bux

So, it's not defined in the standard?

Thomas Tutone wrote:
> Alien wrote:
>
>
> > npos is defined as -1, in both SGI and VS.
> > So npos+1=0?
> > I remember that I read a book saying that npos+1==npos.
>
> The value of npos is implementation defined. On many platforms it is
> -1, but there is no requirement that that be the case. So don't rely
> on it.
>
> Best regards,
>
> Tom

> Thomas Tutone wrote:
> > Alien wrote:
> >
> >
> > > npos is defined as -1, in both SGI and VS.
> > > So npos+1=0?
> > > I remember that I read a book saying that npos+1==npos.
> >
> > The value of npos is implementation defined. On many platforms it is
> > -1, but there is no requirement that that be the case. So don't rely
> > on it.

Alien wrote:
> So, it's not defined in the standard?

Kai-Uwe Bux has proven me wrong. It's defined in std::basic_string as
const size_type npos = -1;

I would have sworn Jossutis said it was implementation-defined.

Sorry to lead you astray.

Best regards,

Tom

Thomas Tutone wrote:
>> Thomas Tutone wrote:
>>> Alien wrote:
>>>
>>>
>>>> npos is defined as -1, in both SGI and VS.
>>>> So npos+1=0?
>>>> I remember that I read a book saying that npos+1==npos.
>>> The value of npos is implementation defined. On many platforms it is
>>> -1, but there is no requirement that that be the case. So don't rely
>>> on it.
>
> Alien wrote:
>> So, it's not defined in the standard?
>
> Kai-Uwe Bux has proven me wrong. It's defined in std::basic_string as
> const size_type npos = -1;
>
> I would have sworn Jossutis said it was implementation-defined.
>

He should have been right, except that it was originaly unspecified, not
implementation defined. Its purpose was to provide a sentinel value,
nothing more. Somehow unspecified got changed to -1, and that's led far
too many people to write code based on that particular value.

--

-- Pete

Author of "The Standard C++ Library Extensions: a Tutorial and
Reference." For more information about this book, see
www.petebecker.com/tr1book.