Re: [earlywarning] Review of draft-rosen-atoca-cap

["Thomson, Martin" <[e-mail]>]

Thanks for taking the time Hannes,

I really wanted to highlight what I thought needed work.  We're obviously still at a very early stage in this work.

> The first challenge we ran into was the lack of a definition of alerts.
> We tried to put a list of alert types in
> http://tools.ietf.org/html/draft-rosen-atoca-server-discovery-00 (see
> Section 6.1) but faced a lot of resistance on the mailing list (see
> http://www.ietf.org/mail-
> archive/web/earlywarning/current/msg00130.html).
> 
> Before this aspect is addressed it is a bit difficult to say whether
> filters, service URNs, resource lists, etc. are the right approach.

The alternative is tougher, but potentially more robust.  You have a single service with no associated taxonomy: urn:service:warning

The results of a query on this service returns all applicable services for the area.  Each contains a description.  The user chooses the warning services based on these descriptions.

There are costs, but it avoids the sticky taxonomy problem.

Another approach is to allow for services to be identified with proprietary URI identifiers.  These generic identifiers could be discovered using out-of-band means (side of bus, web search).  Then LoST doesn't come into it at all, SIP routing (possibly location-based) does all the hard work.  It's probably not-cool if we don't use LoST though.  

--Martin

...snipped...