Re: [Captive-portals] I-D Action: draft-ietf-capport-api-00.txt

[Martin Thomson <[e-mail]>]

On Tue, Feb 6, 2018 at 12:07 PM, David Bird <[e-mail]> wrote:
>> I'd say that the amount of information provided here is rich enough
>> that you want to talk to a human.  Very few networks permit sending of
>> arbitrary packets to arbitrary hosts and the receipt of similar.  The
>> point is to ensure that you are managing expectations.  If I
>> understand the expression of requirements, the idea is that a UE can
>> use the boolean signal, but this other stuff is better presented on
>> the web page.  If the network starts off in a captive state, then that
>> page will be seen (if there is a human), and maybe not used (if there
>> isn't).
>>
>
> I'm not sure I follow that statement...

I'll try again, because that was a little dense.

You asserted (or implied) that a Boolean value was insufficiently
expressive to convey the range of possible policies that a captive
network might impose.  I asserted that while that is true, whatever
you do will be turned into a go/no-go decision by the UE.  This value
is giving the network provider a direct input to that decision.

I acknowledge that you might conclude that we're back to gaming this
out, but I have heard UE vendors say that they really don't want to
probe.  So if the network says that it's good, I think that they will
save the probing for when the network breaks instead.  But we'll let
Tommy and Lorenzo respond.