malloc() and implicit cast

[arnuld]

I have checked the FAQ: http://c-faq.com/malloc/mallocnocast.html
FAQ discusses a special case when programmer has forgotten to do
#include <stdlib.h>. I am including this header and I am not doing any
explicit cast:

#include <stdlib.h>
enum ARRSIZE { MAXSIZE = 100 };
struct dummy
{
int i;
};
int main( void )
{

char *pc;
struct dummy *ptrDummy;

pc = malloc( MAXSIZE );
ptrDummy=malloc(sizeof(struct dummy));

return 0;
}

============ OUTPUT ============
/home/arnuld/programs/C $ gcc -ansi -pedantic -Wall -Wextra test.c
/home/arnuld/programs/C $ ./a.out
/home/arnuld/programs/C $

malloc(size_t n) returns a void pointer and here in my program, I am
assigning malloc returned pointers to 2 different types and I am not
getting any warnings about <implicit cast>.
It has something to do with C90 ?


--
http://lispmachine.wordpress.com/

[Morris Dovey]

arnuld wrote:

I don't understand. Do you think there /should/ be a problem?

--
Morris Dovey
DeSoto Solar
DeSoto, Iowa USA
http://www.iedu.com/DeSoto/

[Stephen Sprunk]

"arnuld" <ul**@kullu.comwrote in message
news:pa****************************@kullu.com...

>I have checked the FAQ: http://c-faq.com/malloc/mallocnocast.html

FAQ discusses a special case when programmer has forgotten to do
#include <stdlib.h>. I am including this header and I am not doing any
explicit cast:

There is no such thing as an "explicit cast". There are implicit and
explicit conversions; the latter uses a cast, and the former does not.

#include <stdlib.h>

....

char *pc;
struct dummy *ptrDummy;

pc = malloc( MAXSIZE );
ptrDummy=malloc(sizeof(struct dummy));

....

malloc(size_t n) returns a void pointer and here in my program, I am
assigning malloc returned pointers to 2 different types and I am not
getting any warnings about <implicit cast>.

There is no such thing as an "implicit cast". There are implicit and
explicit conversions; the latter uses a cast, and the former does not.

Second, a warning is only expected when you _don't_ include the proper
header and you _don't_ use a cast. Since you're including the proper
header, there is no reason for a warning.

What's the problem?

S

--
Stephen Sprunk "God does not play dice." --Albert Einstein
CCIE #3723 "God is an inveterate gambler, and He throws the
K5SSS dice at every possible opportunity." --Stephen Hawking

[Andrey Tarasevich]

arnuld wrote:

I have checked the FAQ: http://c-faq.com/malloc/mallocnocast.html

OK.

FAQ discusses a special case when programmer has forgotten to do
#include <stdlib.h>. I am including this header and I am not doing any
explicit cast:

Yes, that how it should be, if you really read the FAQ: include the header,
don't use the cast.

malloc(size_t n) returns a void pointer and here in my program, I am
assigning malloc returned pointers to 2 different types and I am not
getting any warnings about <implicit cast>.

Of course, you don't. In C language 'void*' pointers are implicitly convertible
to and from other pointer types. What warnings did you expect and why?

--
Best regards,
Andrey Tarasevich

[Jack Klein]

On Tue, 15 Apr 2008 08:13:43 -0700, Andrey Tarasevich
<an**************@hotmail.comwrote in comp.lang.c:

arnuld wrote:

I have checked the FAQ: http://c-faq.com/malloc/mallocnocast.html

OK.

FAQ discusses a special case when programmer has forgotten to do
#include <stdlib.h>. I am including this header and I am not doing any
explicit cast:

Yes, that how it should be, if you really read the FAQ: include the header,
don't use the cast.

malloc(size_t n) returns a void pointer and here in my program, I am
assigning malloc returned pointers to 2 different types and I am not
getting any warnings about <implicit cast>.

Of course, you don't. In C language 'void*' pointers are implicitly convertible
to and from other pointer types. What warnings did you expect and why?

To and from other object pointer type. There is no defined conversion
between pointers to functions and pointers to object types, even
incomplete object types like void.

--
Jack Klein
Home: http://JK-Technology.Com
FAQs for
comp.lang.c http://c-faq.com/
comp.lang.c++ http://www.parashift.com/c++-faq-lite/
alt.comp.lang.learn.c-c++
http://www.club.cc.cmu.edu/~ajo/docs/FAQ-acllc.html

[Peter Nilsson]

Jack Klein wrote:

Andrey Tarasevich <an**************@hotmail.comwrote:

arnuld wrote:

I have checked the FAQ:
http://c-faq.com/malloc/mallocnocast.html
FAQ discusses a special case when programmer has
forgotten to do #include <stdlib.h>. I am including
this header and I am not doing any explicit cast:

Yes, that how it should be, if you really read the
FAQ: include the header, don't use the cast.

I _really_ read the FAQ and took notice of the parenthetical
comment at the end. It highlights that the real issue lies
with using unprototyped functions. Sensible programmers
will use compilers that advise of such things. [Of course,
conforming C90 compilers are not required to issue
diagnostics, but nevertheless the issue has been around
long enough that you'll be hard pressed to find any
conforming compiler that isn't capable of alerting you
to the use of an unprototyped function. Personally, I
think a C programmer be insane not to use that feature
if it was available.]

malloc(size_t n) returns a void pointer and here
in my program, I am assigning malloc returned
pointers to 2 different types and I am not
getting any warnings about <implicit cast>.

Of course, you don't. In C language 'void*' pointers
are implicitly convertible to and from other pointer
types. What warnings did you expect and why?

To and from other object pointer type. There is no
defined conversion between pointers to functions and
pointers to object types, even incomplete object types
like void.

Except for the case of null pointer constants.

--
Peter

[arnuld]

On Tue, 15 Apr 2008 21:00:17 -0500, Jack Klein wrote:

>On Tue, 15 Apr 2008 08:13:43 -0700, Andrey Tarasevich
Of course, you don't. In C language 'void*' pointers are implicitly
convertible to and from other pointer types. What warnings did you
expect and why?

To and from other object pointer type. There is no defined conversion
between pointers to functions and pointers to object types, even
incomplete object types like void.

so an int* is implicitly converted to a void* which then can be
implicitly converted to char* without any warning at all.

--
http://lispmachine.wordpress.com/

find my email ID at the above address.

[arnuld]

On Tue, 15 Apr 2008 21:00:17 -0500, Jack Klein wrote:

To and from other object pointer type. There is no defined conversion
between pointers to functions and pointers to object types, even
incomplete object types like void.

so I conlcude:

1.) Function Pointers: pointer to function returning an int can be
implicitly converted into char* without any any warning message.

2.) Pointers to object types: compiler can implicitly convert and int*
into char* without any warning message to the programmer. C doe snot
require any warning in this case

Is that what you mean ?


--
http://lispmachine.wordpress.com/

[Peter Nilsson]

arnuld wrote:

Jack Klein wrote:

Andrey Tarasevich:

Of course, you don't. In C language 'void*' pointers are implicitly
convertible to and from other pointer types. What warnings did you
expect and why?

To and from other object pointer type. There is no defined conversion
between pointers to functions and pointers to object types, even
incomplete object types like void.

so an int* is implicitly converted to a void*

What int*?

which then can be implicitly converted to char* without any
warning at all.

Consider...

int *ip = malloc(N * sizeof *ip);

The malloc function knows nothing about the type being allocated.
It returns a void * to a region suitably aligned for any object.

There is an implicit conversion from void * to int * in the assignment
of the void * to ip, but as you say there is (generally) no warning.
Nor should you expect there to be one. The implicit conversion of
void * to and from other object or incomplete types is a language
_feature_. [Not necessarily a good one, but a deliberate feature
nonetheless.]

--
Peter

[Richard Heathfield]

arnuld said:

>On Tue, 15 Apr 2008 21:00:17 -0500, Jack Klein wrote:

>To and from other object pointer type. There is no defined conversion
between pointers to functions and pointers to object types, even
incomplete object types like void.

so I conlcude:

1.) Function Pointers: pointer to function returning an int can be
implicitly converted into char* without any any warning message.

I don't see why you conclude this from what Jack said, because it simply
isn't true. A pointer to function, of no matter what return type, cannot
be implicitly converted into *any* other type, let alone a char *.

2.) Pointers to object types: compiler can implicitly convert and int*
into char* without any warning message to the programmer.

No, there is no implicit conversion between int * and char *.

C [does not] require any warning in this case

Implementations are required to diagnose an attempt to assign int * to char
* and vice versa.

Is that what you mean ?

I doubt it, because it's completely wrong.

--
Richard Heathfield <http://www.cpax.org.uk>
Email: -http://www. +rjh@
Google users: <http://www.cpax.org.uk/prg/writings/googly.php>
"Usenet is a strange place" - dmr 29 July 1999

[Peter Nilsson]

Richard Heathfield wrote:

...A pointer to function, of no matter what return type, cannot
be implicitly converted into *any* other type, let alone a char *.

Actually, they can be implicitily converted in limited cases...

int main(int argc, char **argv)
{
int (*fp)() = main; /* okay */
return 0;
}

--
Peter

[Richard Heathfield]

Peter Nilsson said:

Richard Heathfield wrote:

>...A pointer to function, of no matter what return type, cannot
be implicitly converted into *any* other type, let alone a char *.

Actually, they can be implicitily converted in limited cases...

int main(int argc, char **argv)
{
int (*fp)() = main; /* okay */
return 0;
}

ANY dogmatic assertion, no matter how obviously correct, has a
counter-example that disproves it. :-)

In this case, you're dodging round the type system by missing out the
parameter list from fp's type - and yes, that is a valid demonstration
that the type system for function pointers isn't quite as bullet-proof as
I'd suggested. (That doesn't mean that it's a good idea, of course, to do
such dodgery - but it can be a very present help in time of trouble).

--
Richard Heathfield <http://www.cpax.org.uk>
Email: -http://www. +rjh@
Google users: <http://www.cpax.org.uk/prg/writings/googly.php>
"Usenet is a strange place" - dmr 29 July 1999

[CBFalconer]

arnuld wrote:

>Jack Klein wrote:

.... snip ...

>>
To and from other object pointer type. There is no defined
conversion between pointers to functions and pointers to object
types, even incomplete object types like void.

so an int* is implicitly converted to a void* which then can be
implicitly converted to char* without any warning at all.

No, that is a programming error. That particular void* can be
converted back into an int* without loss, but other conversions
(implicit or not) are NOT guaranteed.

--
[mail]: Chuck F (cbfalconer at maineline dot net)
[page]: <http://cbfalconer.home.att.net>
Try the download section.

** Posted from http://www.teranews.com **

[Eric Sosman]

CBFalconer wrote:

arnuld wrote:

>>Jack Klein wrote:

... snip ...

>>To and from other object pointer type. There is no defined
conversion between pointers to functions and pointers to object
types, even incomplete object types like void.

so an int* is implicitly converted to a void* which then can be
implicitly converted to char* without any warning at all.

No, that is a programming error. That particular void* can be
converted back into an int* without loss, but other conversions
(implicit or not) are NOT guaranteed.

In the particular case of thing* to or from char* it's
not ipso facto an error: C specifically permits accessing
an object's representation as an array of char.

A better example might be double* to void* to int*,
which is either an error or at best a dubious practice.

--
Er*********@sun.com

[Andrey Tarasevich]

arnuld wrote:

>

>To and from other object pointer type. There is no defined conversion
between pointers to functions and pointers to object types, even
incomplete object types like void.

so I conlcude:

1.) Function Pointers: pointer to function returning an int can be
implicitly converted into char* without any any warning message.

??? I don't know where you draw that "conclusion" from. Function pointers
_cannot_ be implicitly converted to object pointers. That's what has been said
so far. How you ended up concluding the opposite is beyond me.

2.) Pointers to object types: compiler can implicitly convert and int*
into char* without any warning message to the programmer. C doe snot
require any warning in this case

Compilers _cannot_ implicitly convert 'int* ' into 'char*' directly, meaning
that an attempt to do so would require a diagnostic.

What can be done, is that you can perform that conversion without using an
explicit cast, as a two-step process involving 'void*' as an intermediate type
'int*' -'void*' -'char*'.

int* pi;
...
void* pv = pi; /* implicit conversion 1 */
char* pc = pv; /* implicit conversion 2 */

Needless to say, due to the two-step structure of the process, in general case
the compiler cannot catch the problem here, if there's one. So, of course, it is
not surprising there's no requirement for any diagnostic.

--
Best regards,
Andrey Tarasevich

[Antoninus Twink]

On 16 Apr 2008 at 14:37, Eric Sosman wrote:

In the particular case of thing* to or from char* it's
not ipso facto an error: C specifically permits accessing
an object's representation as an array of char.

A better example might be double* to void* to int*,
which is either an error or at best a dubious practice.

I don't see why isn't necessarily dubious. Portability isn't everything,
and in speed-critical sections of code, using bit-twiddling to
manipulate floating-point numbers can be a very valuable practice.

[CBFalconer]

Eric Sosman wrote:

CBFalconer wrote:

>arnuld wrote:

>>>Jack Klein wrote:

... snip ...

>>>
To and from other object pointer type. There is no defined
conversion between pointers to functions and pointers to object
types, even incomplete object types like void.

so an int* is implicitly converted to a void* which then can be
implicitly converted to char* without any warning at all.

No, that is a programming error. That particular void* can be
converted back into an int* without loss, but other conversions
(implicit or not) are NOT guaranteed.

In the particular case of thing* to or from char* it's
not ipso facto an error: C specifically permits accessing
an object's representation as an array of char.

However having converted to char* there is no GUARANTEE that the
original int* can be recovered from the char*, although it probably
can be. The point is that such games require especial care and
treatment to avoid sneaky insects.

--
[mail]: Chuck F (cbfalconer at maineline dot net)
[page]: <http://cbfalconer.home.att.net>
Try the download section.
** Posted from http://www.teranews.com **

[Eric Sosman]

CBFalconer wrote:

Eric Sosman wrote:

>CBFalconer wrote:

>>arnuld wrote:
Jack Klein wrote:
>
... snip ...
To and from other object pointer type. There is no defined
conversion between pointers to functions and pointers to object
types, even incomplete object types like void.
so an int* is implicitly converted to a void* which then can be
implicitly converted to char* without any warning at all.
No, that is a programming error. That particular void* can be
converted back into an int* without loss, but other conversions
(implicit or not) are NOT guaranteed.

In the particular case of thing* to or from char* it's
not ipso facto an error: C specifically permits accessing
an object's representation as an array of char.

However having converted to char* there is no GUARANTEE that the
original int* can be recovered from the char*, although it probably
can be.

The guarantee is in section 6.3.2.3, paragraph 7:

"A pointer to an object or incomplete type may be
converted to a pointer to a different object or
incomplete type."

So the conversion from int* to char* is allowed.

"If the resulting pointer is not correctly aligned
for the pointed-to type, the behavior is undefined."

Not a problem, since the alignment of char is the weakest
there is.

"Otherwise, when converted back again, the result
shall compare equal to the original pointer."

There's the guarantee.

The point is that such games require especial care and
treatment to avoid sneaky insects.

Here we agree. IMHO, C programs would be better if type-
punning were more difficult -- not impossible, just not so easy
that people can do it without thinking.

--
Er*********@sun.com