Question about (double *)NULL

[Rouben Rostamian]

Umfpack is a C library for computations dealing with sparse
matrices. Several examples in the User's Guide use a certain
cast that puzzles me. Here is an example.

The prototype of the function umfpack_di_symb olic() is:

int umfpack_di_symb olic
(
int n_row,
int n_col,
const int Ap [ ],
const int Ai [ ],
const double Ax [ ],
void **Symbolic,
const double Control [UMFPACK_CONTROL],
double Info [UMFPACK_INFO]
);

The header file umfpack.h has:
#define UMFPACK_CONTROL 20

Here is a sample usage:

int main (void)
{
double *null = (double *) NULL ;
...
(void) umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, null, null) ;
...
return (0) ;
}

(Code fragments are put here by "cut-and-paste"ing from the manual.)

My question is: Does the cast in (double *)NULL accomplish anything?
It seems to me that the call to umfpack_di_symb olic() equivalent to:

umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, NULL, NULL) ;

Comments?

--
Rouben Rostamian

[jacob navia]

Rouben Rostamian a écrit :

Umfpack is a C library for computations dealing with sparse
matrices. Several examples in the User's Guide use a certain
cast that puzzles me. Here is an example.

The prototype of the function umfpack_di_symb olic() is:

int umfpack_di_symb olic
(
int n_row,
int n_col,
const int Ap [ ],
const int Ai [ ],
const double Ax [ ],
void **Symbolic,
const double Control [UMFPACK_CONTROL],
double Info [UMFPACK_INFO]
);

The header file umfpack.h has:
#define UMFPACK_CONTROL 20

Here is a sample usage:

int main (void)
{
double *null = (double *) NULL ;
...
(void) umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, null, null) ;
...
return (0) ;
}

(Code fragments are put here by "cut-and-paste"ing from the manual.)

My question is: Does the cast in (double *)NULL accomplish anything?
It seems to me that the call to umfpack_di_symb olic() equivalent to:

umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, NULL, NULL) ;

Comments?

You are right. The only difference is that you waste code space in a
useless assignment, and that afterwards you write NULL in lowercase.

A
#define null NULL
would accomplish the same thing

[Michael Mair]

Rouben Rostamian schrieb:

Umfpack is a C library for computations dealing with sparse
matrices. Several examples in the User's Guide use a certain
cast that puzzles me. Here is an example.

The prototype of the function umfpack_di_symb olic() is:

int umfpack_di_symb olic
(
int n_row,
int n_col,
const int Ap [ ],
const int Ai [ ],
const double Ax [ ],
void **Symbolic,
const double Control [UMFPACK_CONTROL],
double Info [UMFPACK_INFO]
);

The header file umfpack.h has:
#define UMFPACK_CONTROL 20

Here is a sample usage:

int main (void)
{
double *null = (double *) NULL ;
...
(void) umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, null, null) ;
...
return (0) ;
}

(Code fragments are put here by "cut-and-paste"ing from the manual.)

My question is: Does the cast in (double *)NULL accomplish anything?
It seems to me that the call to umfpack_di_symb olic() equivalent to:

umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, NULL, NULL) ;

Comments?

Educated guess:
It proactively shuts up a C++ compiler abused to compile C...

,---
$ cat doublecast.c
int main (void)
{
double *foo = (void *)0;

return 0;
}

$ gcc -std=c89 -pedantic -Wall -O doublecast.c -c
doublecast.c: In function `main':
doublecast.c:3: warning: unused variable `foo'

$ g++ -std=c++98 -pedantic -Wall -O doublecast.c -c
doublecast.c: In function `int main()':
doublecast.c:3: error: invalid conversion from `void*' to `double*'
doublecast.c:3: warning: unused variable 'foo'
`---
<OT>The correct C++ solution is to use "0" instead of "NULL".</OT>

As most library headers nowadays test whether C or C++ is the
language mode they are processed with, you usually have NULL
defined to be 0 for C++ and (void *)0 for C.
Cheers
Michael
--
E-Mail: Mine is an /at/ gmx /dot/ de address.

[A. Bolmarcich]

On 2006-04-30, Rouben Rostamian <ro****@pc18.ma th.umbc.edu> wrote:

Umfpack is a C library for computations dealing with sparse
matrices. Several examples in the User's Guide use a certain
cast that puzzles me. Here is an example.

The prototype of the function umfpack_di_symb olic() is:

int umfpack_di_symb olic
(
int n_row,
int n_col,
const int Ap [ ],
const int Ai [ ],
const double Ax [ ],
void **Symbolic,
const double Control [UMFPACK_CONTROL],
double Info [UMFPACK_INFO]
);

The header file umfpack.h has:
#define UMFPACK_CONTROL 20

Here is a sample usage:

int main (void)
{
double *null = (double *) NULL ;
...
(void) umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, null, null) ;
...
return (0) ;
}

(Code fragments are put here by "cut-and-paste"ing from the manual.)

My question is: Does the cast in (double *)NULL accomplish anything?
It seems to me that the call to umfpack_di_symb olic() equivalent to:

umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, NULL, NULL) ;

Comments?

The cast is not needed in C (but is in C++).

If the package was originally written before C had function prototypes,
then the sample usage (with arguments of type double *) was needed for
compilers that used different representations for pointers to different
types. With function prototypes, the compiler, if necessary, converts
arguments with a value of NULL to the null pointer of the type of the
corresponding function parameter.

[Jack Klein]

On Sun, 30 Apr 2006 17:28:50 -0000, "A. Bolmarcich"
<ag*****@earl-grey.cloud9.net > wrote in comp.lang.c:

On 2006-04-30, Rouben Rostamian <ro****@pc18.ma th.umbc.edu> wrote:

Umfpack is a C library for computations dealing with sparse
matrices. Several examples in the User's Guide use a certain
cast that puzzles me. Here is an example.

The prototype of the function umfpack_di_symb olic() is:

int umfpack_di_symb olic
(
int n_row,
int n_col,
const int Ap [ ],
const int Ai [ ],
const double Ax [ ],
void **Symbolic,
const double Control [UMFPACK_CONTROL],
double Info [UMFPACK_INFO]
);

The header file umfpack.h has:
#define UMFPACK_CONTROL 20

Here is a sample usage:

int main (void)
{
double *null = (double *) NULL ;
...
(void) umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, null, null) ;
...
return (0) ;
}

(Code fragments are put here by "cut-and-paste"ing from the manual.)

My question is: Does the cast in (double *)NULL accomplish anything?
It seems to me that the call to umfpack_di_symb olic() equivalent to:

umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, NULL, NULL) ;

Comments?

The cast is not needed in C (but is in C++).

No it is not.

<OT>
The C++ standard requires that the macro NULL be defined as an integer
constant expression that evaluates to a value of 0. It does not allow
it to be defined as such an expression cast to void *, as C does.
</OT>

--
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.l earn.c-c++
http://www.contrib.andrew.cmu.edu/~a...FAQ-acllc.html

[Peter Nilsson]

Rouben Rostamian wrote:

Umfpack is a C library for computations dealing with sparse
matrices. Several examples in the User's Guide use a certain
cast that puzzles me. Here is an example.

The prototype of the function umfpack_di_symb olic() is:

int umfpack_di_symb olic
(
int n_row,
int n_col,
const int Ap [ ],
const int Ai [ ],
const double Ax [ ],
void **Symbolic,
const double Control [UMFPACK_CONTROL],
double Info [UMFPACK_INFO]
);

The header file umfpack.h has:
#define UMFPACK_CONTROL 20

Here is a sample usage:

int main (void)
{
double *null = (double *) NULL ;
...
(void) umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, null, null) ;
...
return (0) ;
}

My question is: Does the cast in (double *)NULL accomplish anything?
No.
It seems to me that the call to umfpack_di_symb olic() equivalent to:

umfpack_di_symb olic (n, n, Ap, Ai, Ax, &Symbolic, NULL, NULL) ;

Comments?

It's the same if umfpack_di_symb olic is actually prototyped before its
use.
If the function is not prototyped, then under C90 it will have an
implicit
declaration of...

int umfpack_di_symb olic();

The absense of a prototype would mean that NULL and (double *) NULL
might be different types and/or have different representations when
passed
as an argument. In either case, such a call would invoke undefined
behaviour.

With a prototype in scope the conversion from NULL to (double *) NULL
is implicit.

C99 removed implicit function declarations, but it did not add the
requirement
that a named function be prototyped before use.

Sensible programmers will turn on the 'require function prototypes'
option
on their compilers (if the compiler has one).

--
Peter