On 1/19/18 2:36 PM, Michael Richardson wrote:
I anticipate some not-yet-well-defined situations where the RFC7710 field can not automatically be passed to a browser that has a human there.
We (the ART ADs) have already had conversations where we've talked people out of speculatively staking out space in the .well-known namespace. The guidance here is: come back when you have a clearly defined use-case.
For diagnostics and support service interactions.
Can you expand on this? I mean, if a service provider wants to put things in a consistent place to make diagnostics easier, that's their prerogative. In fact, that's kind of the whole *point* of avoiding synthesized URLs: if we have normative requirements that put certain endpoints in places we spell out, then it takes that flexibility away from operators. The use of .well-known demonstrably makes this situation worse, not better, for running a network.
/a