Re: [cobalt-users] MX record question / Raq4i

[flash22@xxxxxxx]

On Wed, 19 Dec 2001, Ben Liddicott wrote:

> RFC1033 isn't an authority document; as it says at the top, it is a set of guidelines.

I Think you misunderstand what i mean by 'authority', i do not mean
authority in the sense of legal force, or any ability to require, rather,
in the sense of 'having expert knowledge' as in , a professor of chemistry
would be considered an authority on chemistry'

> RFC1480 "does not specify an internet standard".

None of them do in the typical meaning of 'standards', it is a consensus
opinion about how the protocols should be implimented based on how
applications are designed, and resoloutions of issues that have turned up
during implimentation. RFC1480 is intended as a guide and clarification to
to issues that have caused problems, in this sense it represents
'expert opinion', at least in comparison to random opinions from people
taking wild guesses about how they wish the protocol worked as opposed to
how the protocol is generally agreed to work.

> RFC974 "presents a description of how mail systems on the Internet are 
expected to route messages", which bears no relation to how
> they actually do so.

Granted, many implimentations of application software do not properly
follow the recommendations, but is it wise to encourage further
divergence? Do you have any basis for complaint when you do something
incompatable with the specified methods of implimentation and it doesn't
work properly? The only guarantees that come with the RFC's is that they
will result in consistant implimentations if applications consistantly
follow the recommended protocols.

> It thinks for example that hosts use WKS records to show 
>that they accept mail. Does any MTA check WKS records?

You are taking the RFC's out of context, i am fully aware that *parts* of
that RFC are obsoleted by an later RFC that recommends abandoning the WKS
idea because noone bothered to impliment it. That too represents a consus,
and perhaps in that case is a response to having issued an RFC that was in
fact not widely considered to be well thought out, and probably was not
really in agreement with general opinion, unfortunatly, this does happen
from time to time, it's part of the price for having an 'experimental'
protocol ;P

Nonetheless, other parts of that RFC are still applicable to the
implimentation , as the behaviour is not contradicted by later RFC's
 
> Are there any actual standards or standard documents? 
>It would be nice to know.

Yes, there are, but what do you mean by a standard? The things usually
classified as 'standards' are just more formal declarations of consensus,
ANSI is just a collection of representatives from industry that write 
documents that other industries have formally agreed to follow in regards
to implimentation. The RFC's are documents that the internet community has
considered and generaly agreed to follow, how much difference , other than
pure formality, is there? No one can 'enforce' ISO or ANSO 'standards' as
law either really, they just can't claim to be compliant with them if they
are ignoring them.

The IETF does occasionally formalize documents as standards (STDxxx), but
they tend to wait a very long time, as the protocol itself is generally
considered to be somewhat experimental, in flux, and open to improvment
and redesign. This is IMHO one of the best features of the entire
internet, if a better method of doing something is found, and a reasonable
number of people agree it's better, there is no administrative buracracy
dragging it's feet on implimentation...Have you ever looked at how long it
takes the 'official' standards bodies to actually complete a 
standard? ANSI has been known to formalize standards that noone uses
because it took them so long to finish the standard that the technology
involved was totally obsolete by the time they finished them. [I'll pick
on ANSI cause it's in the US , but ISO is no shining example either;] 

(ANSI is also responsible for the Y2K broken Mag tape standard fiasco ;)

The flip side is, if consensus exists as to current implimentation, even
if it's only a lack of further comments to a Request ForComments, then it
is in fact as close to a standard as it's likely to get, so implimentation
is either going to be compatable with the described implimentation, or it
isn't, and if it isn't , you have little cause to complain about it,
unless you are going to propose your own completly independent solution to
a problem, in which case, feel free to write an RFC and gather consensus:)

ok, stepping off soapbox ;P

gsh

[..]