VanLUG Email Archive
> I have a very reliable and knowledgeble friend who has apparently determined
> that NT (now I don't know what version!) sets the ``do not fragment'' bit on
> outgoing datagrams.
That would be normal for a machine that does path MTU discovery.
However, if communication is broken over that link, I would expect
something else has gone wrong with the path MTU discovery. (If it
doesn't get back an ICMP destination unreachable error, or anything
else, it should consider it a black hole and revert to old TCP
behaviour, though.)
> Presumably, NT can't deal with incoming
> fragmented packets either...
As far as I know, it does deal just fine with fragmented packets.
> > Time moves on, so should any OS... If it's to anquard to maintain backward
> > compability, don't... Progress...
Uh, you're talking not just about not maintaining backward
compatability with Linux, but not maintaining backward compatability
with hundreds of millions of devices around the world, and not
conforming to the IPv4 standards. You may feel you can just chuck
standards out the window like that, but I doubt many users would
appreciate it when they start talking to a host through a path that
fragments datagrams.
cjs
Curt Sampson
Info at http://www.portal.ca/
Internet Portal Services, Inc. Through infinite mist, software reverberates
Vancouver, BC (604) 257-9400 In code possess'd of invisible folly.