Is it correct to call string::substr(npos+1)?

[Alien]

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.

 

[Thomas Tutone]

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

 

[Kai-Uwe Bux]

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

 

[Alien]

So, it's not defined in the standard?

Alien wrote:



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]

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

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

 

[Pete Becker]

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


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.