On 22/05/2010, at 8:08 PM, Robert Kisteleki wrote:

> On 2010.05.21. 23:19, Geoff Huston wrote:
>>>> I agree with Terry on this one. I'd personally be much happier with
>>>> verified/unverified instead of valid/invalid. These terms are much
>>>> closer to what we really mean.
>>> 
>>> Ack. We will make this terminology change in the next revision of the
>>> draft.
>> 
>> I disagree with this terminology change - there are three states that are
>> potential outcomes of the process, not two and the proposed terminology
>> does not accommodate this. I request that no change be made in
>> terminology.
> 
> Geoff, you misunderstood. We proposed varified/unverified/unknown instead of 
> valid/invalid/unknown.

I don't believe so at all. My objection was to point out I see insufficient 
semantic distinction between 'unverified' and 'unknown' - 'unverified' says to 
me "did not verify" as distinct from the more precise statement that "I have in 
my possession objects that explicitly 'include' the prefix being tested, and 
these objects have failed to validate the route", or, more succinctly, 
"verification failed". I feel that the appropriate adjective for such routes 
that have failed an active test (as distinct from not have any relevant RPKI 
objects at hand that can be tested against the route) is "invalid".

An alternate terminology that I can see as also making an appropriate 
distinction between failure of validation tests as distinct from not being able 
to test at all can be found in draft-bates-bgp4-nlri-orig-verif-00.txt. There 
the terms are "Authenticated", "Authentication Failed" and "Unauthenticated"

If one were to use "verify" as the base term, then the appropriate distinction 
is "Verified", "Verification Failed", and "Unverified" respectively.

I trust this longer explanation is clearer for you.

And yes, my objection to the proposed terminology change still stands.


  Geoff




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

Reply via email to