Re: [tcpm] About the urgent pointer...

[Fernando Gont <fernando at gont.com.ar>]

Hello, David,

> So, the question is *not* what do the RFCs say, because they are clear. 
> The question is, what do we do about the mismatch between what the RFCs
> say, and what is generally implemented?

One important thing to note is that it is impossible to tell (without
any additional mechanism) what semantics for the UP a TCP implements.
This means that if were were to try to get real stacks implement the
semantics for the UP as specified in RFC 1122, we would basically break
the urgent mechanism.

My understanding from the talks that I had with a few folks at the last
IETF is that the only sensible thing to do is to change the spec
(considering that whether the UP points to the last byte of urgent data
vs. the byte following that chunk, does not really matter -- as long as
all stacks do the same... which is already the case).

Nevertheles, my personal feeling is that regardless of which specific
outcome, we agree that tcpm wg should do something on this issue (and
probably the last posts on this subject are an indication that at least
is clarification is needed).

That said, I believe it would be a good thing to poll the wg to adopt
this I-D -- if anything, to have a starting point to work on this issue.
(FWIW, this last revision has been substantially revised to address the
feedback I got from you and Joe at the last IETF, and feedback from
Alfred that I got while he was helping (a lot) with a related document).

Thanks!

Kind regards,
-- 
Fernando Gont
e-mail: fernando at gont.com.ar || fgont at acm.org
PGP Fingerprint: 7809 84F5 322E 45C7 F1C9 3945 96EE A9EF D076 FFF1