"Magnus Naeslund(f)" wrote:
>
> From: <raybry@timesn.com>
> > Without digging through the archives to see if this has already
> > been suggested (if so, I apologize), why can't the following be done:
> >
> > min(x,y) = ({typeof((x)) __x=(x), __y=(y); (__x < __y) ? __x : __y})
> >
> > That gets you the correct "evaluate the args once" semantics and gives
> > you control over typing (the comparison is done in the type of the
> > first argument) and we don't have to change a zillion drivers.
> >
> > (typeof() is a gcc extension.)
> >
>
> But then again, how do you know it's the type of x we want, maybe we want
> type of y, that is and signed char (not an int like x).
> Talk about hidden buffer overflow stuff :)
What about this then:
#define min(x,y) ({typeof(x) __x=(x); typeof(y) __y=(y); (__x < __y) ?
__x : __y})
This is guaranteed to work the same as the old min/max in all cases but
without side effects. You can still force the comparison to be done with
a certain type by casting the arguments first:
#define typed_min(type, x, y) (min((type)(x), (type)(y)))
...although, if the type used for the comparison is so critical you
maybe shouldn't be hiding it in a macro anyway.
Andrew
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Aug 23 2001 - 21:01:03 EST