Re: iSCSI: Login Request error

[Luben Tuikov <luben@splentec.com>]

Dear Pat,

You are not mentioning anything new, see this message
dated Apr 22, 2002:
http://www.pdl.cmu.edu/mailinglists/ips/mail/msg09751.html

Short message straight to the point is often more helpful.

Long time ago I envisoned there be a simple negotiation
procedure representable as a finite automata,
one for the target and another for the initiator,
both complementing each other (i.e. no contradictions).

I've referred in my past posts about this as ``decision
trees'' -- but no one paid attention.

Furthermore, I'd like those automata (describing the
negotiation process) to be deterministic -- i.e.
simple and no MAY's --- that would make them easily
predictable.

Regarging your post: thank you for agreeing, even in
parenthesis, that there is INCONSISTENCY in the draft.
This was the _main_ point, not quoting the draft.

Please also note that exactly the SAME problem exists
in section 4.2.2. (which you quote as solution).
And then you mention that there would be a loop (A-ha!).
Which is what I proved in my earlier post (to which you
replied). As you can see the logic (contradictory)
in both sections is the _same_.

Anyway, I've mentioned loops before, but no one
paid attention.

This message from Apr 22, 2002, cc to you from me,
http://www.pdl.cmu.edu/mailinglists/ips/mail/msg09751.html
shows the loop in negotiations, and mentions
decision trees (I've mentioned them elsewhere as well).

Anyway, the main point which has to be resolved
is inconsitencies (contradictions) in the draft.

Regars,
Luben

P.S. Another thing, which I've brought up before:
Using context free grammar (EBNF) for
the values of the login/text operational keys.
Q: if someone volunteers their time, would this
be accepted in the draft? Julian?