Hello,

This is a meta discussion on the way we work.
So just exposing some ideas, if they could interest anyone.
Feel free to participate in you want, here or in private.

You may have seen this email on the DNSOP working group:
https://mailarchive.ietf.org/arch/msg/dnsop/gQ0qju2MPDBz4R5KkRYMKL_ioeg
which was prompted by the "Camel" presentation of the DNS at the previous IETF
meeting, and similar interests in other working groups.
The TLS WG seems to have gone into the same kind of questions/directions.

Many points boil down to: not everything that can be done should be done.
(and the future is difficult to predict, hence thinking of being able right at 
the beginning to cover all future use cases, specially those not 
documented/voiced, can be tricky at least or completely ending in the wrong 
direction at worst).

And I think part of the low participation is caused by the following.

Sometimes this working group receives drafts that clearly caters for a specific 
registry or registrar. This is all fine per se and the fact of discussing 
things in public should be welcomed and appreciated. However it may be 
difficult for the other participants to really understand the use case, as this 
may not be always very clear. So you either let it go, or try to understand it, 
or try to offer comments on the proposition to make it more generic, which 
could be completely going an opposite route from the initial needs.

Should every EPP extension used by only one registry be standardized through an 
RFC? Especially a "Proposed Standard" one? I am not sure about that.

It is good of course that an extension is documented and even follows some 
wireframe. This is the purpose of the EPP extensions registry I think, and 
should be enough for some cases. The guidance of the working group would be, 
besides the EPP experts review on the document to advise if this is of generic 
interest or not, so that documents really coming into the WG as adopted 
documents and later WGLC on them and such are really made in such a way they 
are deemed to be useful for the EPP community at large.

For anything that is supposed to benefit the greater good I think that the 
Implementation Section should be mandatory, to be done before a WGLC, with at 
least 2 separate server implementations and 2 separate client implementations.
(and ideally at least one client implementation capable to speak to the 2 
servers ones). It is only when implementations start to exist that the finest 
details of the design can be tested, so their existence should be mandatory 
before pushing a document towards the "Proposed Standard" state.
The corrolary being that the writers will have to find registries and 
registrars willing to implement their proposition, which is one way to show 
interest and consensus.

It would help also I think if the concerned registries speak up and support the 
specific work. I know that we are all under a "one participant one voice" rule, 
but it is clear too that work is done for specific registries or registrars, or 
group of them (like gTLDs).
And for the same reason if any registry thinks that some proposition goes 
completely in the other direction from what they do or plan to do, and even if 
they may not be required to implement the extension, it should be good if they 
speak up. It could help reshape the design to make sure it accomodate more 
cases.

-- 
  Patrick Mevzek

_______________________________________________
regext mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/regext

Reply via email to