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

[ih] Who owns old RFCs ?

Actually, there *was* a formal certification process for the TCP/IP
protocols.? It was established by NIST (formerly NBS) in the early 80s,
when the TCP protocols were being established as DoD Standards.

I know this from personal experience, since I set up at BBN one of the
official labs accredited by NIST to perform the conformance tests.?
There weren't more than a few such labs; it wasn't exactly a growth
business opportunity so it didn't attract much interest.? Probably the
usual players in that world had never heard of TCP and didn't have a
clue how to go about testing it.

The market for the testing service was established when the government
made TCP/IP a DoD standard, and required it as part of many procurement
contracts for military systems.? So the big defense contractors had to
get their systems certified as part of their contract deliverable.?? We
did quite a few such tests for various big-gun contractors, and often
sold additional engineering time to help them understand exactly what
TCP was all about.

There were a few interesting glitches in the overall testing scheme.??
For example, all military systems being procured were required to
implement TCP, and demonstrate it as certified.? But that didn't mean
that TCP actually had to be used as part of their product operation.??

It reminded me of the mid-50s when as a kid I got my very own
12-transistor radio -- only to later learn as I got into electronics
that only 2 of the transistors were actually wired in to the
circuitry.?? Similarly, TCP was in the software, but might not be used
at all.

Anyway, there *was* formal conformance testing, managed by NBS/NIST, in
the early 80s timeframe.? I've always wondered how that happened and who
pushed it through the NBS/NIST machinery.

I suspect the IETF, and NBS/NIST, never knew or never cared to continue
including testing as part of future work on Internet technology.


On 4/24/20 12:46 PM, Dave Crocker via Internet-history wrote:
> On 4/24/2020 4:43 AM, Andrew G. Malis via Internet-history wrote:
>> Some other SDOs and industry fora create consensus conformance test
>> plans
> Historically, the fact that there has NOT been formal 'certification'
> for Internet implementations has been a major feature.?
> Interoperability testing demonstrates that the implementation actually
> works. Comformance testing doesn't.
> With the usual caveats about poor memory, I'll add...
> There was a wonderful panel, around 1990, including Vint and a very
> vocal OSI advocate from Boeing.? Vint commented about our doing
> interops rather than conformance testing and the benefit of
> demonstrating actual ability to... interoperate.? The Boeing advocate
> was a few speakers down the line, so it was some minutes before she
> spoke.
> Her comments included the firm declaration that interoperability
> wasn't possible without formal conformance testing.? In perhaps the
> most overt public act I've ever seen Vint make, he slowly leaned
> slightly forward and slowly turned his head to look at her.
> d/