Re: [lisp] [Technical Errata Reported] RFC7052 (4256)

["Isidoros Kouvelas (kouvelas)" <[e-mail]>]

On Mar 9, 2015, at 19:19, Brian Haberman <[e-mail]> wrote:

> Hi Gregg,
> 
> On 3/9/15 1:09 PM, Gregg Schudel (gschudel) wrote:
>> 
>> On Mar 9, 2015, at 7:07 AM, Brian Haberman <[e-mail]>
>> wrote:
>> 
>>> 
>>> On 2/6/15 12:58 PM, Gregg Schudel (gschudel) wrote:
>>>> 
>>>> On Feb 6, 2015, at 7:44 AM, Brian Haberman
>>>> <[e-mail]> wrote:
>>>> 
>>>>> It turns out that the discussion during review was with the
>>>>> upper limit. I can't find any reference to discussions on the
>>>>> lower limit.  Authors?
>>>> 
>>>> going back to check Brian - it was a while ago.
>>> 
>>> Have we gotten anywhere with this check?  The erratum is sitting in
>>> a less-than-useful state at this point.
>>> 
>>> Regards, Brian
>> 
>> Hi Brian I’m really sorry about droppng this.
>> 
>> I’ve checked internal notes, authors, etc. I cannot find any record
>> of (nor do we recall) us discussing “minimum length.” Looking at it
>> now, the value of “5” in RFC7052 does not match any “real minimum
>> length” that we can see.
>> 
>> Further, (in discussing with Isidor), the ability to return
>> unspecified as an empty address, hence “0” length, makes sense.
>> 
>> So, again - the errate makes sense. No evidence is available to
>> indicate otherwise.
>> 
> 
> Isidor and I discussed whether changing all the lower bounds of 5 to 0
> made sense or if there was a smaller subset of variables that should be
> changed.
> 
> Isidor insisted on lispEidRegistrationLastRegisterSenderLength being
> changed, but there are potentially others.
> 
> I do not want to see a slew of errata roll in over the years changing
> one variable's lower limit at a time.  Keeping in mind, a change for one
> variable actually requires a change to two variables (the affected
> variable and its corresponding Length variable).
> 
> Can the WG identify all the MIB variables that it wants changed with
> this erratum?

The reason for which I think lispEidRegistrationLastRegisterSender is the only use of LispAddressType that needs to be changed is that its access is "read-only". All other uses are "not-accessible” and part of the index of tables.

thanks
Isidor