sizeof operator

[JS]

I read in K&R page 204 that sizeof use on a char returns 1. But when I write
the following I get 4!

int main(void){
printf("%d\n",s izeof('g'));

}

Was it not supposed to return 1?

[Ben Pfaff]

Mark Odell <od*******@hotm ail.com> writes:

Alex Fraser wrote:

printf("%d\n", (int)sizeof c);

Don't you mean:
printf("%u\n", sizeof c);

No, the result of sizeof has type size_t, which may not be type
unsigned.
--
int main(void){char p[]="ABCDEFGHIJKLM NOPQRSTUVWXYZab cdefghijklmnopq rstuvwxyz.\
\n",*q="kl BIcNBFr.NKEzjwC IxNJC";int i=sizeof p/2;char *strchr();int putchar(\
);while(*q){i+= strchr(p,*q++)-p;if(i>=(int)si zeof p)i-=sizeof p-1;putchar(p[i]\
);}return 0;}

[Nils Weller]

In article <d1**********@c hessie.cirr.com >, Christopher Benson-Manica wrote:

Ben Pfaff <bl*@cs.stanfor d.edu> spoke thus:

In C, character constants have type `int', so sizeof 'g' is equal
to sizeof(int).

(semi OT) Why did the authors of the standard make this arguably
non-intuitive stipulation?

Because you may be able to put more than one character into character
constants (but the actual number that fits into it is implementation-
defined any may be 1.)

putchar('foo');

This is only in ANSI/ISO C for compatibility with traditional C, and I
don't think anybody uses this feature anymore. The value of a multi-char
character constant is implementation-defined, so it is not portable at
all (gcc chooses to ignore all but the last character in such a
constant, so the outcome of 'foo' would be 'o'. In addition, gcc always
issues a diagnostic for multi-char character constants because it is so
unlikely that you really meant to use one.)

--
My real email address is ``nils<at>gnuli nux<dot>nl''

[Martin Ambuhl]

Mark Odell wrote:

Alex Fraser wrote:

No, a character in single quotes, as in 'g', is a constant of type signed
int in C (I think this is different in C++).

Try:

int main(void) {
char c;
printf("%d\n", (int)sizeof c);

Don't you mean:
printf("%u\n", sizeof c);

No, since there is no reason to think that a size_t is an unsigned int.
The "correct" specifier for an uncast size_t might be "%lu" or "%llu".

[Ben Pfaff]

Nils Weller <[e-mail] > writes:

In article <d1**********@c hessie.cirr.com >, Christopher Benson-Manica wrote:

Ben Pfaff <bl*@cs.stanfor d.edu> spoke thus:

In C, character constants have type `int', so sizeof 'g' is equal
to sizeof(int).

(semi OT) Why did the authors of the standard make this arguably
non-intuitive stipulation?

Because you may be able to put more than one character into character
constants (but the actual number that fits into it is implementation-
defined any may be 1.)

That is not a cause, that is an effect. If a character constant
had type `char', then by definition you could only fit one `char'
into it.
--
int main(void){char p[]="ABCDEFGHIJKLM NOPQRSTUVWXYZab cdefghijklmnopq rstuvwxyz.\
\n",*q="kl BIcNBFr.NKEzjwC IxNJC";int i=sizeof p/2;char *strchr();int putchar(\
);while(*q){i+= strchr(p,*q++)-p;if(i>=(int)si zeof p)i-=sizeof p-1;putchar(p[i]\
);}return 0;}

[Steven K. Mariner]

Hey, Ben, long time no see.

Your sig block program -- I can't compile it and I can't figure out
why.

I get:

D:\Steve\mcc>mi ngw32-gcc -c sigc0.c -o sigc0.o
sigc0.c: In function `main':
sigc0.c:1: stray '\255' in program

I broke it out to multiline format:

int main(void){
char p[]="ABCDEFGHIJKLM NOPQRSTUVWXY*Za bcdefghijklmnop qrstuvwxyz.
\n",*q="kl BIcNBFr.NKEzjwC IxNJC";

int i=sizeof p/2;

char *strchr();

int putchar();
while(*q) {
i+=strchr(p,*q+ +)-*p;
if(i>=(int)size of p)i-=sizeof p-1;
putchar(p[i]);
}
return 0;
}

And it now reports the problem on line 10:

D:\Steve\mcc>mi ngw32-gcc -c sigc.c -o sigc.o
sigc.c: In function `main':
sigc.c:10: stray '\255' in program

Commenting out line 10 does, in fact permit it to compile.

I've dumped out the source file in hex, ascii and decimal, no 255
character in the source code.

What gives?

[Nils Weller]

In article <87************ @benpfaff.org>, Ben Pfaff wrote:

Nils Weller <[e-mail] > writes:

In article <d1**********@c hessie.cirr.com >, Christopher Benson-Manica wrote:

Ben Pfaff <bl*@cs.stanfor d.edu> spoke thus:

In C, character constants have type `int', so sizeof 'g' is equal
to sizeof(int).

(semi OT) Why did the authors of the standard make this arguably
non-intuitive stipulation?

Because you may be able to put more than one character into character
constants (but the actual number that fits into it is implementation-
defined any may be 1.)

That is not a cause, that is an effect. If a character constant
had type `char', then by definition you could only fit one `char'
into it.

I thought this is a newsgroup about the C programming language, and not
about linguistics. The intent of what I said is obvious to anyone who
does not intentionally misinterpret it. But okay, I'll rephrase:

Because character constants were designed to be capable of carrying more
more than one character (and it follows, obviously, that a ``char'' is
not enough to carry more than one character - hence the need for a
larger data type.)

--
My real email address is ``nils<at>gnuli nux<dot>nl''

[Ben Pfaff]

"Steven K. Mariner" <ma*******@eart hlink.net> writes:

Your sig block program -- I can't compile it and I can't figure out
why.

D:\Steve\mcc>mi ngw32-gcc -c sigc0.c -o sigc0.o
sigc0.c: In function `main':
sigc0.c:1: stray '\255' in program

I think your compiler must be buggy.
--
"I ran it on my DeathStation 9000 and demons flew out of my nose." --Kaz

[Ben Pfaff]

Nils Weller <[e-mail] > writes:

Because character constants were designed to be capable of carrying more
more than one character (and it follows, obviously, that a ``char'' is
not enough to carry more than one character - hence the need for a
larger data type.)

Okay--so, then, *why* were character constants designed to be
capable of carrying more than one character?
--
"...deficie nt support can be a virtue.
It keeps the amateurs off."
--Bjarne Stroustrup

[Jens.Toerring]

Steven K. Mariner <ma*******@eart hlink.net> wrote:

Hey, Ben, long time no see. Your sig block program -- I can't compile it and I can't figure out
why. I get: D:\Steve\mcc>mi ngw32-gcc -c sigc0.c -o sigc0.o
sigc0.c: In function `main':
sigc0.c:1: stray '\255' in program I broke it out to multiline format: int main(void){
char p[]="ABCDEFGHIJKLM NOPQRSTUVWXY*Za bcdefghijklmnop qrstuvwxyz.
\n",*q="kl BIcNBFr.NKEzjwC IxNJC"; int i=sizeof p/2; char *strchr(); int putchar();
while(*q) {
i+=strchr(p,*q+ +)-*p;
if(i>=(int)size of p)i-=sizeof p-1;
putchar(p[i]);
}
return 0;
}
You have managed to get some strange character into it. On the line

i+=strchr(p,*q+ +)-*p;

the character directly in front of the 'p' at the end seems to be the
culprit - it wasn't in Bens signature and is some non-ASCII character
(it arrives here as 0xAD, looking like a '-' but that might be an arte-
fact of the newsreader). But also the line
char p[]="ABCDEFGHIJKLM NOPQRSTUVWXY*Za bcdefghijklmnop qrstuvwxyz.
\n",*q="kl BIcNBFr.NKEzjwC IxNJC";

is dangerous (the space before the '\n' might get lost and and you
would need a '\' at the end of the first line), better make that

char p[]="ABCDEFGHIJKLM NOPQRSTUVWXY*Za bcdefghijklmnop qrstuvwxyz. \n",
*q="kl BIcNBFr.NKEzjwC IxNJC";

Regards, Jens
--
\ Jens Thoms Toerring ___ Je***********@p hysik.fu-berlin.de
\______________ ____________ http://www.toerring.de

[Nils Weller]

In article <87************ @benpfaff.org>, Ben Pfaff wrote:

Nils Weller <[e-mail] > writes:

Because character constants were designed to be capable of carrying more
more than one character (and it follows, obviously, that a ``char'' is
not enough to carry more than one character - hence the need for a
larger data type.)

Okay--so, then, *why* were character constants designed to be
capable of carrying more than one character?

That I don't know. I'd guess that it was intended to automize bit
shifting for you. The value of a multi-char character constant is now
implementation-defined, but historically you really got all chars you
asked for:

'foo'

would turn into

'f' << (CHAR_BIT * 2) | 'o' << CHAR_BIT | 'o'

This may seem more plausible for values written in hexadecimal notation:

'\xff\x12'

would turn into

0xff << CHAR_BIT | 0x12

--
My real email address is ``nils<at>gnuli nux<dot>nl''