Velocity Reviews (http://www.velocityreviews.com/forums/index.php)
-   C++ (http://www.velocityreviews.com/forums/f39-c.html)
-   -   static_cast<unsigned> (http://www.velocityreviews.com/forums/t291596-static_cast-unsigned.html)

 bonham 06-08-2005 03:12 AM

static_cast<unsigned>

i m a beginner in C++. I have some questions:
what is static_cast<unsigned>?
i saw someone use static_cast<unsigned> and static_cast<unsigned char>
together. why?
e.g. static_cast<unsigned>(static_cast<unsigned char>(c))
(c is a char).

Thank you!

Bon

 Me 06-08-2005 03:30 AM

Re: static_cast<unsigned>

> i m a beginner in C++. I have some questions:
> what is static_cast<unsigned>?
> i saw someone use static_cast<unsigned> and static_cast<unsigned char>
> together. why?
> e.g. static_cast<unsigned>(static_cast<unsigned char>(c))
> (c is a char).

Take an implementation where char ranges from -128 to 127, the cast
from char to unsigned char (on this implementation) converts it to the
range 0 to 255 and then the cast to unsigned int leaves it in that 0 to
255 range. Without the unsigned char cast there, converting the char to
an unsigned integer directly (lets say unsigned int ranges from 0 to
0xFFFF on this implementation) would convert it to either 0 to 127 if c
is positive or 0xFFFF+1-128 to 0xFFFF+1-1 if c is negative (all these
ranges here are inclusive).

So basically, it's because char isn't guaranteed to be signed or
unsigned, so we want to force it to become unsigned.

If the code is doing what I think it's trying to do, it's wrong in
general (but works on the majority of 2s complement computers, which is
why you see it in the wild). The correct way is static_cast<const
unsigned char&>(c) because you want to cast the bits of the character
and not the mathematical value of it.

 bonham 06-08-2005 03:51 AM

Re: static_cast<unsigned>

I assume the "&" is there to ensure the cast of bits of character than
the mathematics value in static_cast<const unsigned char&>(c) rite?
why need the "const"?

 bonham 06-08-2005 03:51 AM

Re: static_cast<unsigned>

I assume the "&" is there to ensure the cast of bits of character than
the mathematics value in static_cast<const unsigned char&>(c) rite?
why need the "const"?

 Me 06-08-2005 04:22 AM

Re: static_cast<unsigned>

> I assume the "&" is there to ensure the cast of bits of character than
> the mathematics value in static_cast<const unsigned char&>(c) rite?
> why need the "const"?

static_cast<const unsigned char&>(c) is the same thing (with added
semantics) as *static_cast<const unsigned char*>(&c) so you can see
you're basically intepreting memory a different way. The const is there
for two reasons 1. const safety
(http://www.parashift.com/c++-faq-lit...rrectness.html) and 2.
to allow for binding an r-value to a reference
(http://cpptips.hyperformix.com/cpptips/nconst_tmp_ref3)

 Me 06-08-2005 06:13 AM

Re: static_cast<unsigned>

> char c;
> static_cast<const unsigned char&>(c)

Doh, strike that, that code shouldn't compile (I'm way too used to
VC6's broken compiler). The correct way is to get the unsigned bits is:

*static_cast<const unsigned char*>(static_cast<const void*>(&c)) or
*(const unsigned char*)(const void*)(&c).

Some people just do reinterpret_cast<unsigned char&>(c) or even just
(unsigned char&)(c). The C++ standard doesn't guarantee it works (the C
one does however). Personally, I always cast through void* just to play
it safe.

 bonham 06-08-2005 07:20 PM

Re: static_cast<unsigned>

Thank you very much!

 sparrow 12-01-2012 04:22 PM