On Tue, Feb 16, 2016 at 9:29 PM, Brian Haberman <brian@xxxxxxxxxxxxxxxxxx> wrote:
> Well, section 4 of draft-ietf-dhc-anonymity-profile-07 says:
>
> The choice between the stateful and stateless scenarios depends on
> flag and prefix options published by the "Router Advertisement"
> messages of local routers, as specified in [RFC4861]. When these
> options enable stateless address configuration hosts using the
> anonymity profile SHOULD choose it over stateful address
> configuration, because stateless configuration requires fewer
> information disclosures than stateful configuration.
>
> That seems pretty close from what you want, at least as far as "stateful DHCPv6" is concerned.
I would agree that the above text covers what I interpret as Lorenzo's
concern.
My concern is not with the intent, it's with the wording.
The business of this draft is to provide guidance to implementers. I am an implementer: I wrote the DHCP client currently used in a host OS, and while said OS does not yet support DHCPv6, I am likely to involved with that as well if/when that happens. So I am squarely in the target audience for this document - but as has become clear from this thread, I did not understand the text correctly, even though I was actually paying attention It took a specific response from Christian and an explanation from a colleague before I actually understood what the implications were for host behaviour. We should try to ensure other implementers do not misunderstand the text like I did, by making it clearer.
Let me suggest text again:
When these options enable stateless address configuration (i.e., when
the A flag in a Prefix Information Option is set to 1) hosts using the
anonymity profile SHOULD perform stateless address configuration
and SHOULD NOT use stateful DHCPv6, because stateless configuration
I don't see how that text is different from the text that's already in the draft, except it actually provides clear guidance. Why not use it?