Clarify "multiple networks to choose between"

[nidhijaju]

From Jen Linkova on the mailing list:

The charter says "The algorithm takes as input a fully-qualified
domain name (FQDN), and results in a single established connection to
a single server IP address on a single network. While the algorithm
could apply to scenarios with multiple networks to choose between or
to use simultaneously, or could deal with pools of multiple
connections, such scenarios are out of scope for the working group
deliverables."

IMHO " multiple networks to choose between" needs clarifying. Does
multihoming, especially "multi-prefix, multi-router single-link"
topology or mPVD scenario are in scope?

[ekinnear]

@furry13 would "multiple network attachments" be any better? I think the intent is to cover all of the examples you gave, in the sense that what we'd be writing down is specific to multiple attempts originating a single source IP address.

[furry13]

would "multiple network attachments" be any better?

Thank you Eric, I think it's definitely more clear and specify that multi-interface multihoming is out of scope, but a single interface multihoming is I guess that would mean that something might to be said in the draft about DNS resolution - like it's performed over the same interface? or do we ignore DNS split horizon? anyway, it's another issue). I'll send a PR for this one.