🚀 go-pugleaf

In article <23357B@erik.naggum.no> enag@ifi.uio.no (Erik Naggum) writes:

[ note: positive proposal precedes ObOsiBashing. Skip to end for cable
 jockey abuse]

-----------------------
Getting back to the subject of the original message, one thing that
might be worth a try is setting up a 'free OSI foundation': under EC
law, an interface specification cannot be copyrighted; it would be
entirely legal to take a copy of the OSI standards, and rewrite the
text in a form that would not be owned by the ISO. Whilst great care
would have to be taken to ensure that errors aren't introduced (they'd
probably swamp proper implementations anyway), certain parts of the
standards, such as ASN.1 PDU descriptions could be lifted verbatim
from the text of the standards under the 'must fit' rule. I'd guess
state machines fall under the same category.

If the standards were to be rewritten, they could benefit greatly from being
translated into English (or some other human tounge). If something might be
ambiguous, then surely it would be better to add a clarification, rather
than write the whole thing in 100% legalese. If the worst comes to the worst,
a summary in predicate logic or denotational semantics would be clearer.
Adding realistic examples would be even better (though this may not be 
possible for the session layer [sorry,sorry - did I mention the War?]

The standards which would benefit most from this, judging from the requests
sent to this group, are ASN.1 and the BER. Anybody think this is worth a 
crack?

--------------------

   This explains why X.400(1984) was such a disaster, and ASN.1 needed ISO
   to clean it up.  The entire experiment should be regarded as CCITT
   coming of age and entering the information age.  Perhaps the 1992

The CCITT has no business whatsoever messing around in anything other than
wire stringing; apart from giving an unfair advantage to a small group of
companies, they're not any good at it. Additionally, the fixed 4-year
cycle makes things much too rigid when major defects are found.

The only OSI related standard I'm involved in is NISO Z39.50 (~= ISO SR '92).
This was a protocol that only really existed on paper, until people decided
to start implementing it, using only ASN.1/BER directly over TCP/IP. Since
then, about seven independent implementations have appeared in about 6 
months; problems and missing essential features disovered during implementation
are being fed back constantly, and new features are being tried out before 
they get standardised. My impression is that when an OSI protocol doesn't
duplicate and existing, functional Internet standard, and when the protocol
can be made lightweight enough to make implementation simple, it has a good
chance of success. IMHO X.400 and X.500 do not meet these criteria.


   Erik Naggum             |  ISO  8879 SGML     |      +47 295 0313
			   |  ISO 10744 HyTime   |
   <erik@naggum.no>        |  ISO 10646 UCS      |      Memento, terrigena.
   <enag@ifi.uio.no>       |  ISO  9899 C        |      Memento, vita brevis.

						I remember, I am born of Earth
						I remember, life is short
				[ anybody whose Latin is better than mine
				  managed to translate the whole song?] 


Simon
--
-----
Guest account at TMC 	|	Just Another  WAIS Hacker | DOD# 0612