RE: Urgent as Framing Hint?

["Douglas Otis" <dotis@sanlight.net>]

Matt,

If you wish to change TCP into something defined as 'tightly coupled' then
document this 'coupling' so that those wishing to discuss the 'benefits' can
at least understand what changes are being made.  I fear there are those who
do not understand the impact of your proposal.

> Doug, stop your #$%@! nit picking:
>
> Douglas Otis wrote:
>
> > Perhaps you would take time to explain how the last sentence
> "The result is
> > that there WILL be a TCP segment with a valid TCP pointer
> (urgent flag set)
> > pointing to the FIRST byte of an iSCSI message IN the TCP
> segment." is not
> > mandating change to TCP.
>
> *I* wrote that section of the iSCSI draft.  By the word "WILL" (as you put
> it,) I meant "There will EVENTUALLY be...".  And, as was stated
> many times on
> the IPS reflector, it was with a TCP tightly coupled with the iSCSI
> application.  There was not intent or desire to change the TCP
> interface to
> other applications.

In a proposal you should understand the meaning of will.  It does not mean
may or occasionally and adding the word eventually is still not a correct
statement.  You are now suggesting there is to be a different 'tightly
coupled' TCP used to accomplish this transport.  In other words, not TCP.
This change is a good reason to discuss this proposal within the end2end
reflector. (I did not create the cross posting.)  Those within the IPS WG
are not tasked for this TCP change and though you are proud of this idea, it
is not a good solution.  I have not suggested changes to TCP as I do not
think such modification to be a good idea.  Yes, I have embraced SCTP for
the same reasons you wish to change TCP.  Please, do not hold that against
me.

> Now, I threw that proposal out to try and solve the problem that
> iSCSI wants
> to solve.  Instead of helping to solve the problem, you just
> condemn any idea
> that pops up and throw out SCTP as the panacea that solves all problems,
> instead of "hmm, maybe this idea would work..."
>
> I am glad I put the urgent pointer proposal out there, because others have
> pointed out how there may be problems with using it.  I still believe that
> *if* TCP implementations were implemented correctly, it would work to a
> degree.  You however, have insisted that it was a "modification
> to TCP", when
> in fact, it was never intended to be.

The proposal is clear as to intent.  Your scheme is not at all useful with
standard TCP.

Doug

>
> -Matt Wakeley
> Agilent Technologies
>