Ali Khayam <ali_khayam@yahoo.com> writes:
> Setting the checksum to zero means not performing any
> checksum i.e. not even the header. But, not
> checksumming the header is totally absurd because you
> might end up processing a packet meant for someone
> else.

Not very likely. In fact, it is so unlikely in a modern network that
IPv6 eliminated header checksums entirely, deciding they weren't worth
the bother.

If one insists, though, it is trivial to include a UDP header checksum
in the data portion if you want. It will be static so you only have to
compute it once. You can just shove it in the first couple of bytes of
the data portion of the packet and the receiver could check that it
showed up. Big Deal.

> And UDP Lite is always used in combination with a variation of the
> Link Layer Protocol.

In real networks, you're never going to have control like that over
the link layer.

> UDP Lite has shown a lot of promise for real-time multimedia
> applications on wireless networks.

I'm very active in the IETF's transport area. I hadn't heard of this
previously, and as I said, the functionality seems like for practical
purposes it is already there -- not that I can see how one would
rationally use it.

Unless the IETF were to make this into a standard, which I can nearly
guarantee would never happen, I don't know why we'd bother putting it
into NetBSD.

-- 
Perry E. Metzger		perry@piermont.com
--
"Ask not what your country can force other people to do for you..."