Hi Roni,

responses inline

On Tue, Dec 2, 2008 at 5:38 AM, Roni Even <[EMAIL PROTECTED]> wrote:
> Hi Bruce,
> I am not sure what is covered under
>
> " Addresses concerns about allowing topology-aware implementations to be
> built without changing the base protocol or overlay algorithm." I assume it
> has to do with the symmetric recursive response text.

This is a change to 10.7.3, the stabilization for the chord-like
overlay algorithm.   In -00 the text describes only how to use random
selection for finger table entries and uses MUST to require this.  In
an earlier thread, Das asked if topology-awareness could be used to
select finger table entries, and I agreed to change the wording in
that section to make the random selection a SHOULD but state that an
implementation or additional spec MAY use a different algorithm to
replace the random selection if it meets certain conditions.  New
wording will be in -01.

>
> Just a reminder that we discussed also adding a flag that will not allows
> intermediary peers to keep state of the call.

I have that in my TODO list.  I intend to post a proposed way (or set
of possibilities) to address this in the next few weeks.

I also have the change to make 5.1.2 that defines responsible id,
other id, and private id a separate subsection rather than place it
within the definition of recursive symmetric routing.  I had meant to
include this change on my slide in Minneapolis, but forgot.  Will also
get that done in the next few weeks.

Bruce


> Roni Even
>
> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of
> Bruce Lowekamp
> Sent: Monday, December 01, 2008 7:15 PM
> To: [email protected]
> Subject: [P2PSIP] reload base draft changes from 73
>
> During the presentation on draft-ietf-p2psip-base-00, there was
> consensus to make a number of changes to the draft.  Unless there are
> significant objections raised on list, I will make these changes to
> the draft.  A few other issues will be in separate threads.
>
> Bruce
>
> - remove TUNNEL
>
> In Dublin we agreed to remove TUNNEL if no adequate use cases were
> brought to list to justify its inclusion.  None have been brought to
> list.
>
> - Decouple overlay description from enrollment server
>
> The draft currently says the overlay description is obtained from the
> enrollment server.  Remove that statement so the overlay description
> can come from anywhere and can point to the enrollment server.
>
> - Reword finger table entry selection to allow other algorithms
>
> Addresses concerns about allowing topology-aware implementations to be
> built without changing the base protocol or overlay algorithm.
>
> - Large messages
>
> Error codes will be introduced to signal messages too big.  (more
> changes may be agreed on later, but we appeared to have consensus that
> at least those are necessary in the base.)
> _______________________________________________
> P2PSIP mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/p2psip
>
>
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip

Reply via email to