Re: [DNSOP] Second Working Group Last Call for draft-ietf-dnsop-extended-error

[Petr Špaček <[e-mail]>]

On 24. 10. 19 19:24, Eric Orth wrote:
> On Tue, Oct 22, 2019 at 6:49 AM Tony Finch <[e-mail] <mailto:[e-mail]>> wrote:
> 
>     I expect almost no-one can do anything with EDE without
>     getaddrinfo() EAI_ return code extensions.
> 
>  
> In many cases, especially when DoH is in use, Chrome uses its own built-in stub resolver.  So EDE is certainly a reasonable option for us without any changes to getaddrinfo().
> 
> On Mon, Oct 21, 2019 at 1:48 PM Petr Špaček <[e-mail] <mailto:[e-mail]>> wrote:
> 
>     2. Second problem is that it is uncelar if there is going to be a consumer: Did *anyone* from stub resolvers said a word about this draft? Is it useful as it is? Is there an experimental implementation in stub to consume this information?
>     dnsop has history of tweaks which never get used by stubs, and this draft in particular is very expensive to implement in resolver code.
> 
>  
> Chrome DNS has no specific plans decided for EDE yet.  But we do think it's generally a good idea and are looking forward to it.  I don't have any implementation-ability concerns with the current draft.

Excellent, I'm glad to hear that!

This draft creates a new registry with info-codes, and content of the registry is very geeky, so it seems necessary to somehow map error codes to categories which can be communnicated to users in terms of "a random web browser user".

Would it be possible to get some time allocation of your UX design folks, before we set the protocol in stone?


This draft evolved several times when it comes to error categorization, and I believe we will never know if we (= dnsop geeks) got the categorization right until someone attempts to design UX for it.

History of draft-ietf-dnsop-extended-error:
- version 00 - categorization by FLAG
- version 01 - categorization by FLAG + RCODE
- version 07 - no categorization at all

There were other proposals for categorization which did not appear in draft text:
- Shane Kerr - HTTP-like categorization like 1xx, 2xx, 3xx, ...
https://mailarchive.ietf.org/arch/msg/dnsop/DMtP4XuqN132Jxt8O0BTZPFsokg

- Petr Spacek (me) - dichotomy local (near) error vs. remote (far) end error
https://mailarchive.ietf.org/arch/msg/dnsop/b3wtVj_aWm24PXyHr1M9NMj3LJ0

It would be super helpful if an experienced UX designer can have look at give back feedback:

- Is the current version with "no categorization" okay?
- Will it be okay in future if dnsop starts adding new codes?
- Or should we add some inherent categorization into the protocol to make it future-proof?
- If a categorization is needed, what form is best for UX? (please suggest!)

Thank you very much for your time!

-- 
Petr Špaček  @  CZ.NIC