[ih] Fwd: Gateway Issue: Certification (was Re: booting linux on a 4004)

[John Day]

> Begin forwarded message:
> 
> From: John Day <jeanjour at comcast.net>
> Subject: Re: [ih] Gateway Issue: Certification (was Re: booting linux on a 4004)
> Date: October 3, 2024 at 16:20:46 EDT
> To: Brian E Carpenter <brian.e.carpenter at gmail.com>
> 
> You give the OSI effort far too much credit. There were a lot of things that OSI created opportunities for consultants. But then isn’t the motto of consultants, “There is one born every minute”?
> 
> No one I knew took it seriously and as I said, the conformance stuff was pretty lousy/goosy. It would immediately get back into the debates between the PTTs and everyone else.
> 
> The attempts at formal description were more useful. But even there, the view was that they were ultimately no more correct than the prose specification.  As I said, they were useful in finding bugs during the development process.
> 
> John
> 
>> On Oct 3, 2024, at 16:03, Brian E Carpenter via Internet-history <internet-history at elists.isoc.org> wrote:
>> 
>> There's a variant of Godwin's law something like: "As an online
>> discussion of Internet history grows longer, the probability of a
>> comparison involving OSI approaches 1."
>> 
>> OSI people were very keen on formal conformance testing and
>> certification. It was supposed to be one of the great benefits
>> of formal specifications, state diagrams, and the like. Consultants
>> made good money out of it.
>> 
>> Over here, we had rough consensus and running code.
>> 
>> We know what happened.
>> 
>> But it's true that the IETF has repeatedly failed to solve the
>> problem that Dave identified ("know exactly which RFCs they need
>> to implement"). See the following URL to appreciate the problem:
>> https://github.com/becarpenter/book6/blob/main/20.%20Further%20Reading/RFC%20bibliography.md
>> 
>>   Brian
>> 
>> Regards
>>  Brian Carpenter
>> 
>> On 04-Oct-24 06:54, Dave Crocker via Internet-history wrote:
>>> On 10/3/2024 10:43 AM, Jack Haverty via Internet-history wrote:
>>>> It's curious to me that such mechanisms have not been created for the
>>>> Internet Industry.
>>> Informal interoperability testing, versus formal compliance testing, was
>>> and remains a key distinction between the Internet's culture and the
>>> cultures of various other standards organization. Compliance testing is
>>> typically expensive and incomplete.  (As a tool for initial code
>>> debugging, tests like that can be efficient; as a guarantee of field
>>> interoperability, not so much.)
>>> There was a wonderful panel that Vint was on, circa 1990, along with a
>>> number of other folk, including a vigorous OSI proponent from Boeing.
>>> Vint made his comments about experiences with the Internet's technology
>>> and specifically noted the reliance on interoperability testing rather
>>> than (bench) compliance testing.
>>> Other panelists made various comments and then the Boeing person made
>>> theirs, vigorously asserting that it is not possible to get widespread
>>> interoperability without formal compliance testing.
>>> It was fun to watch Vint slowly lean slightly forward and then slowly
>>> turn his head toward the Boeing person.
>>> d/
>> -- 
>> Internet-history mailing list
>> Internet-history at elists.isoc.org
>> https://elists.isoc.org/mailman/listinfo/internet-history
>