Kent M Pitman <pitman@world.std.com> wrote:
+---------------
| But standards making is NOT about design.
+---------------
(*sigh*) Oh, would that that were true! And indeed it used to be, but...
In the area of networking hardware & protocols, HP-IB (as IEEE-488)
and HIPPI were about the last standards which codified existing,
working practice more-or-less unchanged. Since then, all new networking
hardware/protocols have been "design by standards committee", by what
some cynics call the principle of "maximum mutual disadvantage". That is,
if something is already *working*, well then, the company that made it
work would have an "unfair" competitive advantage if the working system
got standardized, wouldn't it, and we can't have *that*, can we??!?
As a prime example, consider FDDI...
IEEE 802.3 *almost* left D-I-X Ethernet alone, but broke it by requiring
that only 802.2 LLC be used on top. Fortunately, Xerox helped the pragmatists
squeak by that one by relocating a few of the previously-assigned Ethernet
"types" that had the same values as IEEE 802.3 "lengths" [which fell in
the same bit positions on the wire], so that now all of the "types" are
disjoint from the legal values of "length", so classic D-I-X Ethernet can
co-exist on the same wire as 802.3. [*Whew!!* That was close!] And a similar
end-run snuck Ethernet types back into "pure" 802.2 media such as FDDI,
by using the "SNAPpy Xerox" hack (everyone agreeing that an LLC/SNAP frame
with OUI=0 implies that the two "reserved for organizational use" bytes
would be an Ethernet type). [Another close one!]
And so it goes. *All* of the ATM world has been "design by standards
committee". Even the "rebel" AAL5 (a radically simpler alternative
to AAL4) was proposed to the ANSI T1S1.5 before it was ever implemented.
[And don't even get me *started* about ATM flow-control...]
-Rob
-----
Rob Warnock, 7L-551 rpw3@sgi.com http://reality.sgi.com/rpw3/
Silicon Graphics, Inc. Phone: 650-933-1673
2011 N. Shoreline Blvd. FAX: 650-933-4392
Mountain View, CA 94043 PP-ASEL-IA