Multiple Inclusion of Standard Header Files

[David T. Ashley]

Are the standard header files (<stdio.h> for example) always safe to include
more than once?

Is this just good programming practice on the part of the average person who
writes these things, or is this guaranteed in a standard?

 

[Ian Collins]

Are the standard header files (<stdio.h> for example) always safe to include
more than once?
Yes

Is this just good programming practice on the part of the average person who
writes these things, or is this guaranteed in a standard?

It is - 7.1.2 paragraph 4.

 

[Guest]

Are the standard header files (<stdio.h> for example) always safe to include
more than once?

Is this just good programming practice on the part of the average person who
writes these things, or is this guaranteed in a standard?

It's guaranteed, with one exception. 7.1.2p4 states:
"Standard headers may be included in any order; each may be included
more than once in a given scope, with no effect different from being
included only once, except that the effect of including <assert.h>
depends on the deï¬nition of NDEBUG (see 7.2). [...]"

 

[sam_cit]

Are the standard header files (<stdio.h> for example) always safe to include
more than once?

Actually, many header file uses pre-processor statement

#ifndef INCLUDE
#define INCLUDE 1
....
....
....
#endif

which would take care that contents of the header are not replaced
again by the pre-processor...

 

[Richard Heathfield]

(e-mail address removed) said:

Actually, many header file uses pre-processor statement

#ifndef INCLUDE
#define INCLUDE 1
...
...
...
#endif

which would take care that contents of the header are not replaced
again by the pre-processor...

Nor would any other header that used the same identifier. Fortunately, that
isn't how things are done.

You pick a *unique* name, e.g. something like:

#ifndef H_CUSTOMER_HEADER
#define H_CUSTOMER_HEADER 1
....
#endif

That way, you don't disable other headers further down the list.

 

[Malcolm McLean]

(e-mail address removed) said:


Nor would any other header that used the same identifier. Fortunately,
that
isn't how things are done.

You pick a *unique* name, e.g. something like:

#ifndef H_CUSTOMER_HEADER
#define H_CUSTOMER_HEADER 1
...
#endif

That way, you don't disable other headers further down the list.

#ifndef foo_h
#define foo_h
/* foo.h - defines a wonderful function */
double foo(int x);
#endif

No need to define as 1

 

[Richard Heathfield]

Malcolm McLean said:

#ifndef foo_h
#define foo_h
/* foo.h - defines a wonderful function */
double foo(int x);
#endif

No need to define as 1

True enough. I do it anyway, against the possibility that I might one day
find a use for changing it to 2.

 

[Serve Laurijssen]

True enough. I do it anyway, against the possibility that I might one day
find a use for changing it to 2.

Just thinking about it, defining as 1 helps makes it possible to use both

#if !foo_h
and
#ifndef

I never do it though