#ifdef vs #if vs #if defined

[Michael B Allen]

Which is the preferred method for preprocessor tests and why?

#ifdef XYZ
or
#if XYZ
or
#if defined(XYZ)

and

#elif
or
#elsif

and

#ifndef XYZ
or
#if !defined(XYZ)

Mike

[Eric Sosman]

Michael B Allen wrote:

Which is the preferred method for preprocessor tests and why?

#ifdef XYZ
or
#if XYZ
or
#if defined(XYZ)
The first and third are equivalent; use whichever
you please. The second is slighly different from the
other two, since it will test "false" if XYZ is
defined but with the value zero.
and

#elif
or
#elsif
The first is strongly preferred, except by people
who enjoy reading error messages.
and

#ifndef XYZ
or
#if !defined(XYZ)

Equivalent; use whichever you like.

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

[Walter Roberson]

In article <pa********************************@ioplex.com>,
Michael B Allen <mb*****@ioplex.com> wrote:
:Which is the preferred method for preprocessor tests and why?

YMMV, but:

- expression format is the easiest to change later.
- you are likely going to need to use expression format for some of
what you do. It gets to look odd if you deliberately avoid expression
format when possible.

:#ifdef XYZ
:or
:#if XYZ
:or
:#if defined(XYZ)

The second of those does not have the same meaning as the other two.
If XYZ is defined as 0, then #ifdef and #if defined are both true
but the #if XYZ form is false.
--
Pity the poor electron, floating around minding its own business for
billions of years; and then suddenly Bam!! -- annihilated just so
you could read this posting.

[philipx]

Michael B Allen wrote:

Which is the preferred method for preprocessor tests and why?
#ifdef XYZ
or
#if XYZ
or
#if defined(XYZ)
"#ifdef XYZ" is different from "#if XYZ", the former is testing the
existence of XYZ while the latter tests the value of XYZ.
IMO, you can freely choose between "#ifdef XYZ" and "#if defined(XYZ)", with
one exception, you cannot test for more than one thing in a single
"#ifdef".
consider the following, it's better to use "#if defined" than "#ifdef":

/* Test for the support of C99 Standard */
#if defined(__STDC_VERSION__) && __STD_VERSION__ > 199901L
/* ...something legal in C99 only */
#elif
/* ...something alternative */
#endif

#elif
or
#elsif
I can't remember I saw something like #elsif, is it a compiler-specific
extension? (okay, I might saw it when I used Turob C 2.0, but that's long
time ago, 10+ years, I am not quite sure).
by the way, #elsif is not a standard c preprocessor feature, avoid using it.
and
#ifndef XYZ
or
#if !defined(XYZ)
Mike

What's true for "#ifdef" and "#if" is also true here.
--
Java is not platform-independent, it is the platform!

[CBFalconer]

Walter Roberson wrote:

.... snip ... --
Pity the poor electron, floating around minding its own business for
billions of years; and then suddenly Bam!! -- annihilated just so
you could read this posting.

Minor nit - electron-positron annihilation produces gamma rays, not
electricity nor light.

--
"If you want to post a followup via groups.google.com, don't use
the broken "Reply" link at the bottom of the article. Click on
"show options" at the top of the article, then click on the
"Reply" at the bottom of the article headers." - Keith Thompson

[Keith Thompson]

CBFalconer <cb********@yahoo.com> writes:

Walter Roberson wrote:

... snip ...

--
Pity the poor electron, floating around minding its own business for
billions of years; and then suddenly Bam!! -- annihilated just so
you could read this posting.

Minor nit - electron-positron annihilation produces gamma rays, not
electricity nor light.

Minor nit - gamma rays are light. They're just very very very very
very blue.

--
Keith Thompson (The_Other_Keith) ks***@mib.org <http://www.ghoti.net/~kst>
San Diego Supercomputer Center <*> <http://users.sdsc.edu/~kst>
We must do something. This is something. Therefore, we must do this.

[Chris Torek]

In article <X8*****************@read1.cgocable.net>
philipx <ak***@cogeco.ca> wrote:

IMO, you can freely choose between "#ifdef XYZ" and "#if defined(XYZ)", with
one exception, you cannot test for more than one thing in a single
"#ifdef".
Right.

Also, as a side note, the parentheses are not required when using
the "defined" pseudo-keyword (it is really just a magic identifier,
rather than a keyword, since keywords do not exist in these phases
of translation):

#ifdef X
#if defined(X)
#if defined X

are all synonymous.
consider the following, it's better to use "#if defined" than "#ifdef":

/* Test for the support of C99 Standard */
#if defined(__STDC_VERSION__) && __STD_VERSION__ > 199901L
/* ...something legal in C99 only */

This test is a little odd, because if __STDC_VERSION__ is not
defined, the test:

#if __STDC_VERSION__ > 199901L

is entirely legal and "means" the same thing as:

#if 0 > 199901L

which is of course false. In preprocessor expressions, undefined
identifiers *must* be replaced with 0. This is why:

#if sizeof(int) == 4

*must* produce a diagnostic, because -- assuming you have not done
something silly like "#define sizeof" -- it is syntactically the
same as:

#if 0(0) == 4

Once the required diagnostic has been emitted, a C compiler can go
back and "notice" that "sizeof" is not only not-defined but will
also become a valid keyword later, and then figure out that you
want it to act as though sizeof is allowed in preprocessor expressions.
A suitable diagnostic, for a smart compiler that does this, might
be something annoying like:

foo.c, line 123: warning: I handle "sizeof" the way you mean it
here, because "I M SMRT", but ANSI/ISO C compilers are not
required to, so your code may not work on other systems.

Of course, if you do "#define sizeof" -- e.g.:

#define sizeof(x) 3
#if sizeof(int) == 4
#undef sizeof

-- then other conditions apply.
--
In-Real-Life: Chris Torek, Wind River Systems
Salt Lake City, UT, USA (40°39.22'N, 111°50.29'W) [phone]
email: forget about it http://web.torek.net/torek/index.html
Reading email is like searching for food in the garbage, thanks to spammers.