How Not To Test Pointers

Pointer tests in the realm of implementation-dependent behavior

So, what’s wrong with this little snippet of code:

extern void *foo();

void *p = foo(); if( p > 0 ) { printf( “Pointer is okay\n” ); }

When I saw this, my first thought was that I’d never seen that done, nor had I ever imagined anyone doing it.

The compiler apparently liked it just fine, but the behavior was not what the person who wrote that line of code expected - i.e. the conditional evaluated false on a non-null pointer.

According to H&S , the behavior in this case is undefined unless the pointers being compared point into the same array or structure.

Bottom line - don’t do that! The normal four ways to test a pointer always work.

if( p ) if( ! p ) if( p == 0 ) if( p != 0 )

(And, no, this wasn’t my code! :-)