Arnaud Lacombe <lacombar%gmail.com@localhost> writes: > make sense, any objection to fix NULL definition ? > > --- a/sys/sys/null.h > +++ b/sys/sys/null.h > @@ -5,7 +5,7 @@ > #ifndef NULL > #if !defined(__GNUG__) || __GNUG__ < 2 || (__GNUG__ == 2 && > __GNUC_MINOR__ < 90) > #if !defined(__cplusplus) > -#define NULL (void *)0 > +#define NULL ((void *)0) > #else > #define NULL 0 > #endif /* !__cplusplus */ So far I'm not convinced that our previous definition is wrong, or that "sizeof NULL" is valid (in the sense that it must work in all conforming environments). There's a general tradition in macros to parenthesize the whole thing and the arguments so that various usages work, and in this case there's also the issue of dealing with other code that's really out there. So all in all I think adding () is a good step.
Attachment:
pgpVibEjZykTt.pgp
Description: PGP signature