cwr@pnet01.cts.com (Will Rose) (11/15/89)
Rex Jaesche, who was (is?) a member of the ANSI C standards committe,
discussed this macro extensively in the Sep/Oct 88 C Users Journal. His
summary was that ANSI C (in which we *believe*, right?) had three possible
definitions for the NULL pointer constant:
#define NULL 0
#define NULL 0L
#define NULL (void *)0
Most MSDOS C compilers (he goes on to say) conditionally compile NULL into one
of the first two definitions; the third case should work in all
ANSI-conforming programs, but there are plenty of cases where it won't work in
non-ANSI code.
To understand why, read his (six page) article.
Will
-----------------------------------------------------------------------
"If heaven too had passions | Will Rose
even heaven would | UUCP: {nosc ucsd hplabs!hp-sdd}!crash!pnet01!cw
grow old." - Li Ho. | ARPA: crash!pnet01!cwr@nosc.mil
| INET: cwr@pnet01.cts.com
UUCP: {nosc ucsd hplabs!hp-sdd}!crash!pnet01!cwr
ARPA: crash!pnet01!cwr@nosc.mil
INET: cwr@pnet01.cts.comhenry@utzoo.uucp (Henry Spencer) (11/16/89)
In article <710@crash.cts.com> cwr@pnet01.cts.com (Will Rose) writes: > #define NULL 0 > #define NULL 0L > #define NULL (void *)0 > >Most MSDOS C compilers (he goes on to say) conditionally compile NULL into one >of the first two definitions; the third case should work in all >ANSI-conforming programs, but there are plenty of cases where it won't work in >non-ANSI code. Actually, all three definitions *will* work in all ANSI-conforming programs compiled on ANSI-conforming implementations. The choice between them is based solely on the desire to break as few non-conforming programs as possible, by making NULL be a zero of the "right size" (the "void *" choice is there because some machines have pointers longer than "long"). Of course, on some machines there *is* no single "right size", or a zero is not the right bit pattern, in which case non-conforming programs are up the creek.