What is the Q-Method of telnet option implementation?
deostroll
the concept?
--deostroll
David Schwartz
> I've stumbled on the topic. Is there an article online which explains
> the concept?
This is the best explanation. The idea is to avoid a loop when each
side agrees to grudgingly agree to what the other side prefers:
http://www.faqs.org/rfcs/rfc1143.html
The nightmare scenario is this:
1) A sends its preference, "WILL". B sends its preference, "WONT".
2) A gets "WONT" and agrees to accept B's preference, sending "DONT".
B gets "WILL" and agrees to accept A's preference, sending "DO".
3) A gets "DO" and agree to accept B's acceptance of A's preference,
sending "WILL".
B gets "WONT" and agrees to accept A's acceptance of B's preference,
sending "DONT".
4) Goto 2.
It's kind of like two people refusing to walk through a door as each
one says "after you". In each step, each person believes they are
agreeing with the other, but at no step do they actually agree with
each other because each side is agreeing with the other's position
from the previous step.
DS
deostroll
don't understand tht part...
--deostroll
David Schwartz
> It talks about maintaining something for storing replies we sent...I
> don't understand tht part...
The RFC talks about a lot of different methods for solving the
negotiation loop problem at various different levels of detail. It's
confusing if you think that all of the RFC is documenting one method
or miss the transition between methods or levels. It's also easy to
confuse a discussion of a disfavored method with a favored one. The
RFC is not a good example of clear explanation.
My advice to you would be to read from the section that begins
"EXAMPLE STATE MACHINE / FOR THE Q METHOD OF IMPLEMENTING TELNET
OPTION NEGOTIATION". That only documents the recommend mechanism and
does it at the right level of detail for someone who has to code it.
DS