[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ih] Who owns old RFCs ?

On 4/24/20 12:54 PM, Dan Lynch via Internet-history wrote:
> Back in the 80s I created Interop so vendors could demonstrate compliance with the IETF RFC standards. The idea of a testing institute to ensure compliance was floated and found too burdensome by everyone so public demonstrations became the efficient way. Our motto became ?I know it works. I saw it at Interop!?  Of course there was months of voluntary testing at my lab in Sunnyvale that preceded the public demonstrations at Interop. Self interest motivated every one.

We certainly broke a lot of stuff on the Interop Shownet, but that was 
the intent. [Although igniting the lobby of the Las Vegas Convention 
center was more of an accident.? ;-)

The Interop shows created a mighty metronome - everybody had to get 
their act together once or twice a year and show that their stuff played 
nicely with others.

And don't forget the TCP/IP Bakeoffs.? Those were a lot of fun - but 
those were the days when everybody brought their source code and wasn't 
shy about saying "look, this is how we did it".? We even did 'em without 
cover of an NDA, just good manners.

We did similar things at the SIPit events.? I remember one person 
proudly saying "My SIP implementation is invulnerable because it is 
written in Python."? It was written in Python but it was not invulnerable.

For a couple of years we had trouble getting certain companies to 
interoperate - most notably HP - because they insisted on using all of 
that SNAP and other gunk that IEEE had specified on top of DIX (Digital, 
Intel, Xerox) Ethernet that the rest of everybody was using.? They 
insisted that they were "conformant" - but they certainly did not 

One forgotten hammer that drove interoperability was the Air Force's 
ULANA contract - it was a very large procurement effort for its era.? I 
was part of the TRW team (led by David Kaufman and Geoff Baehr) and we 
had to create a body of off-the-shelf products, from routers to PC NICs, 
that all worked together.

In our own testing stuff at IWL we have an allergy to the word 
"conformance".? Our approach is interoperability, with a big heaping 
helping of "will it still work if the other side isn't playing by the 
rules or, if the other side is playing by part of the rules that few 
others use?"

We'd still be in the dark ages if we had to have full "conformance".? 
For instance I know of no IPv4 stack that is fully capable of 
reassembling all possible variations of IP fragmentation.? But nobody 
really needs to do that in real life.

 ??? --karl--