On Sat, Feb 25, 2006 at 11:48:39PM +0100, Johannes Berg wrote:
> On Sat, 2006-02-25 at 17:09 -0500, Jeff Garzik wrote:
> 
> > I strongly ACK the use of rtnetlink, but I leave it to John Linville 
> > (wireless maintainer) to do a full review, and merge...
> 
> The question is just -- do we really want another interface that is
> going to be deprecated in the long run?

        Everything is eventually replaced in the long run. The
question there is how long is the long run...
        It may be a bit too early to make definite statements about
the new Wireless interface, however I have a strong feeling that it
may not support all wireless drivers we have in the kernel, at least
not in its first incarnation. One of the proposal I've seen was
strongly tied to the new 802.11 stack, which is good because it will
make a rich API, but not all driver may be ported to the new 802.11
stack, especially non-802.11 drivers.

> Since we essentially said WE must die in previous threads, adding this
> will just increase the backward compatibility burden (because we can't
> remove the ioctl interface anyway).

        WE-over-RtNetlink is not dependant on WE-over-ioctl, and those
two paths are cleanly separated in the code, so we can remove the
ioctl interface when we want.
        Also, both use the same internal API, so it does not increase
the backward compatibility burden, if you want to support WE in the
kernel you only have to write one backward WE wrapper.

> OTOH, with the WE-over-netlink, maybe we can get rid of WE-ioctl
> earlier, and that helps with the conversion to the yet to be defined new
> interface?

        It's true that going from ioctl to RtNetlink require to
rethink some of the user space code. One example, interfaces are
addressed by ifindex instead of ifname. But to say if it helps or not,
I don't know, we should ask people writing those apps...

> johannes

        Thanks, have fun...

        Jean


-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to