On Thu, Jan 16, 2025 at 2:45 AM Mario Loffredo <mario.loffr...@iit.cnr.it> wrote: > > > Il 15/01/2025 21:05, Andrew Newton (andy) ha scritto: > > On Wed, Jan 15, 2025 at 5:38 AM <kowa...@denic.de> wrote: > >> Hi Andy, > >> > >> On 14.01.25 17:42, Andrew Newton (andy) wrote: > >>> On Fri, Jan 10, 2025 at 12:35 PM <kowa...@denic.de> wrote: > >>>> [PK] as indicated above it is not about the extension specifying the > >>>> referral. > >>> Understood, but how does the current text not allow this? > >>> > >> I was guided by the interpretation that the text in 4.2 applies to > >> referrals specified by a given extension. If the intention of the text > >> is that it applies to any referrals, also to those specified elsewhere, > >> then it's OK, but maybe giving it an explicit wording in the text would > >> make it also very clear to the authors of future extensions. > > I think that is a good enhancement. Along with "servers MUST NOT use > > multiple extensions in a response with processing requirements over > > the same referrals." > > Sorry by I didn't catch this. Does it mean that a server MUST NOT > include multiple extensions in the same link ? >
Going with Pawel's wording: "servers MUST NOT use multiple extensions in a response with processing requirements over the same referrals, which different clients would not be able to process in a deterministic way." What we are trying to say is that if the "foo" extension says that clients should rot13 referrals before dereferencing them and the "bar" extension says that clients should rot26 referrals before dereferencing, that leads to non deterministic behavior and therefore servers must not do that. -andy _______________________________________________ regext mailing list -- regext@ietf.org To unsubscribe send an email to regext-le...@ietf.org
