On Mon, Aug 30, 2004 at 03:46:57PM +1000, Nathan Scott wrote:
> On Sun, Aug 29, 2004 at 06:52:49AM +0100, viro@xxxxxxxxxxxxxxxx wrote:
> > Signed-off-by: Al Viro <viro@xxxxxxxxxxxxxxxxxxxxxxxx>
> This seems like unnecessary code churn to me.
> These macros are used in this way so that some XFS kernel
> headers are the same as the userspace equivalents, so that
> we don't end up maintaining diverged duplicate headers.
> The patch could go further and remove __BYTE_ORDER, not
> sure why that wasn't done. But, whatever, I'd prefer it
> isn't applied unless there's a compelling reason here
> that I've overlooked (eg, it actually fixes something?).
a) #if where #ifdef would be enough
b) use of (legitimate, but *ugly*) semantics of undefined identifiers in #if
(they are replaced with 0, all right, but more often than not it's *not* the
intended behaviour - same story as with assignment in conditional, etc.)
I can kill __BYTE_ORDER and its users - gladly. But that's a separate series
of patches (already have some done).
Anyway, NAK is NAK. Linus, the rest of series is independent from this one,
so just drop this patch - it won't affect others.