On Sun, 25 Oct 1998, David S. Miller wrote:
>
> For example, everybody in the egcs camp just decided that clobbers
> and inputs must not overlap. Nobody told me why, and why they can't
> just be automatically converted to early-clobbers inside gcc.
>
> You're telling the compiler two different things which conflict.
Which has worked before (and was certainly the _only_ way to do it in
gcc-1.40 which I started with).
True, I do remember this problem.
However, now I feel the argument against the construct is even
stronger especially since the documentation explicitly states against
doing it as Berndt has pointed out in another mail.
We could go on in circles about this issue. But I know that if I
presented corollary arguments about older Linux kernels it would not
hold very much water. For example I could construct a similar
argument about the timeval being modified by select() and that
0.99pl15 kernels at the time provided no other way to achieve the
functionality it provided, even though modern kernels do. It used to
work before, and at that time it was certainly the _only_ way to do
it.
This doesn't hold water, because we move on and we can make something
which works without that facility working as "it used to work", so
please understand our position. We want to improve the compiler
similar to how we want to improve the kernel, this involves some
growing pains and some of them require some understanding by both
parties.
Later,
David S. Miller
davem@dm.cobaltmicro.com
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/