Re: [manet] DLEP-17: duplicate functions

I have previously used a Session Update message to carry extension data
items after Session Init.  Rather than include 'unknown' data items too
early (check my dlep destination service announcement draft)

I must admit to not being keen on adding an extra Init step to the state
machine with another message.

There might be some cross-over here however.  If we handle 'unexpected
data item' as a 'continue' status code, rather than a 'terminate', then
the issue disappears (apart from the 2-pass parse, but it's only at
session init, so I don't care about performance here).

Rick (still on vacation...)

On 09/12/15 22:23, Henning Rogge wrote:
> On Wed, Dec 9, 2015 at 11:09 PM, Stan Ratliff <ratliffstan@gmail.com> wrote:
>> Yeah... two-stage parsing... don't know about you, but I'd like to avoid
>> that.
> What I am doing at the moment is having a place to store every data
> item I support... this way its possible to do a single-stage parsing.
>
> Default values are only delivered by the radio to the router if I
> remember it correctly, which happens after the router tells the radio
> what it supports.
>
> This means the router can instantly drop the session if it gets a
> metric it does not support.
>
>>> We only have a few mandatory metrics (link speed related things,
>>> right?)... I am not sure I am good with dropping them too.
>>>
>>>
>>> This gives a router at least a chance to determine what this radio is
>>> useful for.
>>
>> IMO, being able to negatively acknowledge the Session Metrics Init message
>> does the same thing, I think - a router has the opportunity to look at the
>> metrics being advertised by a modem, and make the determination you speak
>> of. Just for the record - I think a modem implementation that doesn't
>> announce link speed is pretty silly. But then, I don't claim to understand
>> each and every possible use case. Something that is totally absurd to us
>> might actually work for someone...
> In fact its even "okay" to report both link speed and maximum link
> speed once at a constant... so this should be EASY for any kind of
> implementation.
>
> Link speed is hard to measure for the router... lets keep it in there.
>
> Henning
>
> _______________________________________________
> manet mailing list
> manet@ietf.org
> https://www.ietf.org/mailman/listinfo/manet
>

Re: [manet] DLEP-17: duplicate functions - Options & extensions negotiations