Re: An IETF repository for working code in our protocols?

[Erik Kline <[e-mail]>]

>
>
> The things I find most useful are the those like the NIST and crfg RFC
>> style of test vectors and expected outputs (and even sometimes expected
>> internal state!).
>>
>> I would love to see some kind of standard approach for documenting what
>> are essentially unit tests: given these inputs/this state/this HTTP method
>> call/... -> expect these outputs/state changes/lists of observable behaviour
>>
>
> Those constructs you mention --- protocol style unit tests --- are
> extremely valuable.  The SIP working group, when it was active, produced a
> few such documents that contained complete protocol flows that aided in
> testing and understanding of how the protocol works.  Many of these
> documents were made into Informational RFCs.  The ALTO working group had
> also produced a HTTP style RESTful API document that outlines the state of
> HTTP headers and payload; alas, it did not progress to an RFC, but it was
> very helpful when we had interoperability events for ALTO.
>

Brilliant!  Yes, this was exactly the kind of thing of which I was thinking.


>
>> Given the centrality of interoperability to the IETF, this seems like
>> something we might be able to explore.
>>
>
> Yes, absolutely.  However, the working code (good working code) being
> available is a bit of an orthogonal, though equally important, discussion
> as is the one for establishing test vectors.  So at least I would not like
> to conflate these two at the current time.
>
>
But surely we'd want the code to path the protocol unit tests, yes?  ;-)