stupid C preprocessor question

stupid C preprocessor question

[Steven G. Kargl]

I learned K&R C several years ago and now have a need
to return to programming in C.  Is the following legal C?

#include <math.h>
double tmp(void) {
   double x;
   x = sin(1.);
#if 0
   Can't do this
#endif
   return x;
}

-- 
Steve
http://troutmask.apl.washington.edu/~kargl/

Re: stupid C preprocessor question

[Zack Weinberg]

"Steven G. Kargl" <kargl@troutmask.apl.washington.edu> writes:

> I learned K&R C several years ago and now have a need
> to return to programming in C.  Is the following legal C?

The only bit of your code that is objectionable is this bit

> #if 0
>    Can't do this
> #endif

which provokes undefined behavior, because of the unmatched single
quote.  Use a real comment (/* Can't do this */) instead.

zw

Re: stupid C preprocessor question

[Steven G. Kargl]

Zack Weinberg wrote:
> "Steven G. Kargl" <kargl@troutmask.apl.washington.edu> writes:
> 
> > I learned K&R C several years ago and now have a need
> > to return to programming in C.  Is the following legal C?
> 
> The only bit of your code that is objectionable is this bit
> 
> > #if 0
> >    Can't do this
> > #endif
> 
> which provokes undefined behavior, because of the unmatched single
> quote.  Use a real comment (/* Can't do this */) instead.
> 

I'm translating a large chunk of Fortran to C.  The use of
#if 0 ... #endif makes it much easier.  I don't have access to
the C99 standard, but Harbison and Steele state in a discussion
of #if ... #endif constructs that "A group of lines that is 
discarded is not processed by the preprocessor." (H&S, 5th ed.
page 62).  

kargl[202] gcc -c a.c
a.c:5:7: missing terminating ' character

It appears that gcc is preprocessing the discarded lines.

-- 
Steve
http://troutmask.apl.washington.edu/~kargl/

Re: stupid C preprocessor question

[Zack Weinberg]

"Steven G. Kargl" <kargl@troutmask.apl.washington.edu> writes:
>
> I'm translating a large chunk of Fortran to C.  The use of
> #if 0 ... #endif makes it much easier. 

I don't see why /* */ don't work for you.

> I don't have access to the C99 standard, but Harbison and Steele
> state in a discussion of #if ... #endif constructs that "A group of
> lines that is discarded is not processed by the preprocessor." (H&S,
> 5th ed.  page 62).
>
> kargl[202] gcc -c a.c
> a.c:5:7: missing terminating ' character
>
> It appears that gcc is preprocessing the discarded lines.

What H&S mean is that if you write

#if 0
#define FOO bar
#endif

FOO is not defined as a macro.  Lines inside a failed #if are still
tokenized (they have to be - how else to find the #endif?

zw

Re: stupid C preprocessor question

[Robert Dewar]

Steven G. Kargl wrote:

> I don't have access to
> the C99 standard, but Harbison and Steele state in a discussion
> of #if ... #endif constructs that "A group of lines that is 
> discarded is not processed by the preprocessor." (H&S, 5th ed.
> page 62).

This is just plain wrong, or rather casually careless.

> It appears that gcc is preprocessing the discarded lines.

The lines discarded still have to be valid sequences of tokens.

Re: stupid C preprocessor question

[Steven G. Kargl]

Robert Dewar wrote:
> Steven G. Kargl wrote:
> 
> > I don't have access to
> > the C99 standard, but Harbison and Steele state in a discussion
> > of #if ... #endif constructs that "A group of lines that is 
> > discarded is not processed by the preprocessor." (H&S, 5th ed.
> > page 62).
> 
> This is just plain wrong, or rather casually careless.
> 

H&S have a correct explanation.  It's my reading comprehesion
after a 14+ hour day seems to be diminished.

-- 
Steve
http://troutmask.apl.washington.edu/~kargl/