On Wed, Apr 15, 2026 at 7:41 AM Sasha Romijn <[email protected]> wrote:

> Hello Christopher,
>
> Thank you for the review of our draft. Responses inline:
>
> On 15 Apr 2026, at 07:09, Christopher Inacio via Datatracker <
> [email protected]> wrote:
> >
> >
> > * As a general comment, I'm concerned about the lack of specificity on
> the JWS
> > cryptography.  IANA has a lot of registered options for JSON Web
> Signatures,
> > algorithms, compression, etc.  Every client implementation has build all
> of
> > those options in order to handle anything the server creates?  Is there a
> > mandatory to implement sub-set of those?
>
> This section was also mentioned in other reviews, and we have refined it
> in:
>
> https://github.com/mxsasha/nrtmv4/commit/a4b1105e48247ce6a64bcbc2b8ecd570b0810e55
>
> We now require ES256, with other options allowed within certain
> constraints, while the responsibility is on the server operator to ensure
> clients actually support this.
>
> [ci] thanks for the updates, I think it makes the document more clear.


> > * Being more specific about the Key, key lifetime, etc. would be useful
> here;
> > that seems a not well defined
>
> We feel this is reasonably addressed in 9.6. If there are still things
> missing, could you elaborate?
>
> [ci] some part of me would like a stronger basis for the lifetime of the
key relative to the amount of data and strength of the crypto used.  I
think that would be extremely hard to do here; so I admit that I might not
be entirely reasonable on this.  It's not blocking and I won't lose any
sleep worrying that you haven't determined empirically that 1-year keys are
strong enough.  That's just the basis of my comment.


> > * I think in the terminology it would be helpful to be very clear that
> > `database` in this context is the collection of IRR snapshot and delta
> files;
> > it does not refer to any type of RDBMS or similar system.  (That is
> fairly
> > clear already, but I don’t think it would hurt to be extra explicit.)
>
> We've clarified this in the terminology.
>
> [ci] looks good to me.


> > * > 260    Version  An incremental number that identifies the IRR
> Database at a
> >> 261       particular point in time.
> >  I think this can be more clear, maybe something like (I’m sure you can
> do
> >  better, but): `Version - A monotonically increasing number that
> identifies a
> >  particular IRR Database data version, scoped by a Session Identifier,
> at a
> >  particular point in time` `Session Identifier - A generated identifier
> which
> >  defines a particular instance of a copy of the IRR database upon which
> >  Snapshot Files, Delta Files, and Notifications are associated`
>
> We have also clarified this in the terminology.
>
> Sasha


I'm happy with the changes.

thanks

-- 
Chris Inacio
Carnegie Mellon University
_______________________________________________
GROW mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to