[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Captive-portals] today's BoF

On Sun, Nov 13, 2016 at 8:36 PM, Erik Kline <[email protected]> wrote:
Notably absent:
- David Bird

Sorry I couldn't make it, I'm not at IETF. 

Everyone familiar with RFC 7710

HTTP code 511, RFC 6585

Mariko Kobayashi volunteers to write industry survey of captive portals!

Abrahamsson: there is lots of overlap with alto working group

Desirable to solve:
- automatic login, with credentials
Sounds like HS2.0 
- time left on the clock / expiration time
Can be signaled with ICMP I-D 
- agree to legal terms? clients cache hashes of known acceptable terms
of service?
This seems out of place for the W.G. I suggest we focus on networking, not application, issues. 
- hotel room number and last name style login?
Same here 
- different categories of service
signaling the existence of additional classes of service (e.g. you are currently at a low QoS, but can go higher?), or? 
- agree on simple authentication mechanism
WISPr? :) 
- data allotment (usage based) 
- query current status (maintain a persistent connection?)
Why query when ICMP informs... 
- what to do when provisioning is done: use as is, teardown and re-associate
    - a la provisioning APNs

Apple and Android watch don't have browsers, so can't login into captive portals

Margaret and Donnelley volunteer to write a document proposing a REST API
- Use 7710
- call REST API on 7710 URI

I think it is a tall order to ask this API to give you a complete picture of your captivity. Will it will return a list of all resources you can freely access within the walled garden, for example? (That would tell the user a lot about your network and partnerships...)  My though is that the station should, for the most part, assume it can access resources until it is told it can't ... learning the bounds of it's captivity on a need to know basis. 
- includes what you need to do to access the portal
- want to include expiration time...

Coova chili has a REST API
- open source captive portal
- should look at this for comparison/...

Schinazi: in France any public wifi requires name, email and terms of service

Some discussion of the ICMP messages.

I believe :) the CP ICMP I-D is a simple signal, with a similar purpose to existing ICMP messages for packet drop signaling, that, together with RFC7710, solves some of the mentioned issues, and a big one not mentioned in the list: fixing the HTTPS "home page" problem. 
Usage-based captive portals.

Barry volunteers to discuss being a co-chair

Captive-portals mailing list
[email protected]