Re: lastVerified: optional vs. extension

[Edward Lewis <edlewis@arin.net>]

At 20:06 -0800 12/5/02, Hong Liu wrote:
>Ed,
>
>Questions for clarification regarding whose opinion
>counts:
>- when you said users of EPP clients, do you mean
>registrars/resellers?

yup

>- when you said users of EPP servers, do you mean
>registries?

yup

>- when you said implementors, do you mean third party
>EPP software vendors?

uh, yeah, but that includes you (and Scott, Joe, Michael, Rick, 
et.al. - I don't mean to omit anyone, but you get the idea).

>- how about other people who do not fall into one of
>the above three categories, but are also concerned
>about this topic, such as regulartory agencies? The
>verification can be done via a third party other than
>the registry and its registrars.

no - they may influence the users, but let's keep this in 
perspective.  We are talking about whether to put last-verified as an 
optional element of the base protocol or in an extension.

>Also, questions for clarification on the metric:
>- are technical issues such as simplicity, continuity,
>and legacy data migrations key considerations?

I can't imagine that these are important here.  (Remember that we are 
just talking about last-verified date.)

>- is lack of established policy framework for the use
>of this element a key consideration?

I think so.  I'd hate to see two servers interpret the date different 
given just one way of expressing it.  If there are two semantics 
involved, two different ways of communicating should be present.

Think about client C and server S1 and server S2.  S1 and S2 disagree 
on last-verified date's meaning and usage.  Is there a need to keep 
the syntax the same?  It may be nice because it makes the parsing 
code the same.  But if you require them to be the same (optional 
elements in the base), you are forcing a decision to be made.  If you 
leave them to extensions, then as new interpretations of the 
last-date are made, syntax can be decided later.

(Also C can verify that it is using the right date semantics by 
looking at the XML - if you debug down to that level.)

-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis                                          +1-703-227-9854
ARIN Research Engineer