Re: [spring] Proposed policy on reporting implementation and interoperability

[Robert Raszuk <[e-mail]>]

Joel,

On #1 IMHO implementations of all MUSTs is what is necessary
for interoperability. If something is not needed for base
protocol functionality it should not be MUST in the first place. After all,
aren't we about assuring interoperable implementations here ?

As to #2 - no need for separate web pages. IETF wiki works fine with a
markup syntax. So all that is needed from any WG is just to setup a single
wiki page - rest is just simple text :)  Alternative would be to keep such
reports on the Wg git.

The problem with including any implementation details in the spec is that
the frequency it gets updated is much higher then spec can handle ..
especially after freezing it after publication.

Thx,
R.

On Sat, Aug 13, 2022 at 12:02 AM Joel Halpern <[e-mail]> wrote:

> With regard to point 1 about MUSTs and implementations, we chose this
> because we recognize the reality that what people say is an implementation
> of an RFC may not include all the MUST clauses.  If we were protocol
> police, that would be a problem.  In this case, we would rather know about
> partial implementations and the fact that they are partial.
>
> As for point 2, the purpose here is to capture the snapshot.  If people
> want to separately maintain web pages, I am sure we can get the pages set
> up to complement this effort.
>
> Yours,
>
> Joel
> On 8/12/2022 6:00 PM, Robert Raszuk wrote:
>
> Hi Joel,
>
> First thank you & AD for initiating this.
>
> Two questions/comments below:
>
> #1:
>
>> 2) Each implementation description MUST include either a statement that
>> all MUST clauses in the draft / RFC are implemented, or a statement as to
>> which ones are not implemented.
>>
>
> How can you allow any implementation to be compliant with an draft/RFC if
> normative MUSTs are not implemented. That is extremely risky if I am
> reading it correctly.
>
> Of course as others pointed out draft may have a lot of optional elements
> which may or may not be implemented at the discretion of the vendor or use
> cases. But I would not extend it for MUSTs.
>
> #2:
>
> > Including the reports in the document is preferred.
>
> As an example in IDR we converged on documenting implementations on IETF
> IDR wiki page. Wouldn't it be nice to have some alignment in this
> method across WGs ? At least within the Routing Area ?
>
> Many thx,
> Robert
>
>