Pointer to array and gcc warnings

[vippstar@gmail.com]

Today I got a confusing message from gcc (I'm aware, those don't break
conformance ;-)

In function 'main':
7: warning: format '%d' expects type 'int', but argument 2 has type
'char (*)[1u]'

The code is

#include <stdio.h>

int main(void) {

char p[3][2];

printf("%d\n", &p[0]); /* this is errornous, but it's on purpose
there to generate the warning */

return 0;
}

I'm curious, what is gcc talking about?
Why is &p[0] a char (*)[1u]?

My first guess was that char (*)[N] pointers get promoted to char (*)
[1] pointers in VLA arguments. (it doesn't make much sense to me
though)

Can anyone explain (don't answer with a yes; actually do so :-) gcc's
behavior here?

[Richard]

Re: Pointer to array and gcc warnings

vippstar@gmail. com writes:

Today I got a confusing message from gcc (I'm aware, those don't break
conformance ;-)
>
In function 'main':
7: warning: format '%d' expects type 'int', but argument 2 has type
'char (*)[1u]'
>
The code is
>
#include <stdio.h>
>
int main(void) {
>
char p[3][2];
>
printf("%d\n", &p[0]); /* this is errornous, but it's on purpose
there to generate the warning */
>
return 0;
}
>
I'm curious, what is gcc talking about?
Why is &p[0] a char (*)[1u]?
>
My first guess was that char (*)[N] pointers get promoted to char (*)
[1] pointers in VLA arguments. (it doesn't make much sense to me
though)
>
Can anyone explain (don't answer with a yes; actually do so :-) gcc's
behavior here?

"This is off topic in c.l.c. We do not know anything about gcc
here. please ask in a newsgroup more appropriate to your question. The
Gnu C newsgroup is down the corridor on the left."

Ah. That felt good.

[Kenny McCormack]

Re: Pointer to array and gcc warnings

In article <gbb32c$aai$3@r egistered.motza rella.org>,
Richard <rgrdev@gmail.c omwrote:

>vippstar@gmail .com writes:
>

>Today I got a confusing message from gcc (I'm aware, those don't break
>conformance ;-)
>>
>In function 'main':
>7: warning: format '%d' expects type 'int', but argument 2 has type
>'char (*)[1u]'
>>
>The code is
>>
>#include <stdio.h>
>>
>int main(void) {
>>
> char p[3][2];
>>
> printf("%d\n", &p[0]); /* this is errornous, but it's on purpose
>there to generate the warning */
>>
> return 0;
>}
>>
>I'm curious, what is gcc talking about?
>Why is &p[0] a char (*)[1u]?
>>
>My first guess was that char (*)[N] pointers get promoted to char (*)
>[1] pointers in VLA arguments. (it doesn't make much sense to me
>though)
>>
>Can anyone explain (don't answer with a yes; actually do so :-) gcc's
>behavior here?

>
>"This is off topic in c.l.c. We do not know anything about gcc
>here. please ask in a newsgroup more appropriate to your question. The
>Gnu C newsgroup is down the corridor on the left."
>
>Ah. That felt good.

Bathe in the glory. There's hope for you (*) yet.

(*) To get accepted into the he-man CLC regs society.

[jameskuyper@verizon.net]

Re: Pointer to array and gcc warnings

vipps...@gmail. com wrote:

Today I got a confusing message from gcc (I'm aware, those don't break
conformance ;-)
>
In function 'main':
7: warning: format '%d' expects type 'int', but argument 2 has type
'char (*)[1u]'
>
The code is
>
#include <stdio.h>
>
int main(void) {
>
char p[3][2];
>
printf("%d\n", &p[0]); /* this is errornous, but it's on purpose
there to generate the warning */
>
return 0;
}
>
I'm curious, what is gcc talking about?
Why is &p[0] a char (*)[1u]?
>
My first guess was that char (*)[N] pointers get promoted to char (*)
[1] pointers in VLA arguments. (it doesn't make much sense to me
though)

VLA normally refers to variable length arrays. You don't have any of
those in this program. I suspect you mean variable arguments. No such
promotion should occur.

As far as the C standard is concerned, the type of &p[0] is char(*)
[2]. Why gcc described it as char(*)[1u] is something you'll have to
take up with gcc. It's definitely version dependent. I have access to
gcc versions 3.2.3 and 3.3, both of which simply say:

warning: int format, pointer arg (arg 2)

[vippstar@gmail.com]

Re: Pointer to array and gcc warnings

On Sep 23, 7:13 pm, jameskuy...@ver izon.net wrote:

vipps...@gmail. com wrote:

Today I got a confusing message from gcc (I'm aware, those don't break
conformance ;-)

>

In function 'main':
7: warning: format '%d' expects type 'int', but argument 2 has type
'char (*)[1u]'

>

The code is

>

#include <stdio.h>

>

int main(void) {

>

char p[3][2];

>

printf("%d\n", &p[0]); /* this is errornous, but it's on purpose
there to generate the warning */

>

return 0;
}

>

I'm curious, what is gcc talking about?
Why is &p[0] a char (*)[1u]?

>

My first guess was that char (*)[N] pointers get promoted to char (*)
[1] pointers in VLA arguments. (it doesn't make much sense to me
though)

>
VLA normally refers to variable length arrays. You don't have any of
those in this program. I suspect you mean variable arguments. No such
promotion should occur.

Whoops, I meant "variadic arguments".

As far as the C standard is concerned, the type of &p[0] is char(*)
[2]. Why gcc described it as char(*)[1u] is something you'll have to
take up with gcc. It's definitely version dependent. I have access to
gcc versions 3.2.3 and 3.3, both of which simply say:
>
warning: int format, pointer arg (arg 2)

Thanks. :-)

[Flash Gordon]

Re: Pointer to array and gcc warnings

vippstar@gmail. com wrote, On 23/09/08 17:16:

On Sep 23, 7:13 pm, jameskuy...@ver izon.net wrote:

>vipps...@gmail .com wrote:

>>Today I got a confusing message from gcc (I'm aware, those don't break
>>conformance ;-)
>>In function 'main':
>>7: warning: format '%d' expects type 'int', but argument 2 has type
>>'char (*)[1u]'
>>The code is
>>#include <stdio.h>
>>int main(void) {
>> char p[3][2];
>> printf("%d\n", &p[0]); /* this is errornous, but it's on purpose
>>there to generate the warning */
>> return 0;
>>}
>>I'm curious, what is gcc talking about?
>>Why is &p[0] a char (*)[1u]?

<snip>

>As far as the C standard is concerned, the type of &p[0] is char(*)
>[2]. Why gcc described it as char(*)[1u] is something you'll have to
>take up with gcc. It's definitely version dependent. I have access to
>gcc versions 3.2.3 and 3.3, both of which simply say:
>>
>warning: int format, pointer arg (arg 2)

>
Thanks. :-)

I suspect it was a bug that has been fixed. I get a far more useful warning:
markg@brenda:~$ gcc -ansi -pedantic -Wall -Wextra -g3 t.c
t.c: In function ‘main’:
t.c:7: warning: format ‘%d’ expects type ‘int’, but argument 2 has type
‘char (*)[2]’
markg@brenda:~$
--
Flash Gordon
If spamming me sent it to smap@spam.cause way.com
If emailing me use my reply-to address
See the comp.lang.c Wiki hosted by me at http://clc-wiki.net/

[vippstar@gmail.com]

Re: Pointer to array and gcc warnings

On Sep 23, 10:06 pm, Flash Gordon <s...@spam.caus eway.comwrote:

vipps...@gmail. com wrote:

Today I got a confusing message from gcc (I'm aware, those don't break
conformance ;-)

<snip>

I suspect it was a bug that has been fixed. I get a far more useful warning:
markg@brenda:~$ gcc -ansi -pedantic -Wall -Wextra -g3 t.c
t.c: In function ‘main’:
t.c:7: warning: format ‘%d’ expects type ‘int’, but argument 2 has type
‘char (*)[2]’
markg@brenda:~$

Ah, thank you too.
For anyone interested, I'm using gcc 4.0.3

[Barry Schwarz]

Re: Pointer to array and gcc warnings

On Tue, 23 Sep 2008 08:40:58 -0700 (PDT), vippstar@gmail. com wrote:

>Today I got a confusing message from gcc (I'm aware, those don't break
>conformance ;-)
>
>In function 'main':
>7: warning: format '%d' expects type 'int', but argument 2 has type
>'char (*)[1u]'
>
>The code is
>
>#include <stdio.h>
>
>int main(void) {
>
char p[3][2];
>
printf("%d\n", &p[0]); /* this is errornous, but it's on purpose
>there to generate the warning */

The & operator always results in a value of type pointer to something.
A pointer value is never the appropriate type of argument for %d. And
undefined behavior does break conformance.

>
return 0;
>}
>
>I'm curious, what is gcc talking about?
>Why is &p[0] a char (*)[1u]?

p is an array of 3 array of 2 char.

p[0] is the first array of 2 char in the above object.

&p[0] is the address of p[0] with type pointer to array of 2 char. The
syntax for this type is char (*)[2].

The standard does not impose any quality requirements on the text of
diagnostics and this diagnostic is not even a required one. So what
you have is a useful message that happens to suffer from a slight
inaccuracy in terminology.

>
>My first guess was that char (*)[N] pointers get promoted to char (*)
>[1] pointers in VLA arguments. (it doesn't make much sense to me
>though)
>
>Can anyone explain (don't answer with a yes; actually do so :-) gcc's
>behavior here?

--
Remove del for email