Viktor Dukhovni wrote in <[email protected]>: |> On 12 Nov 2026, at 06:25, Steffen Nurpmeso <[email protected]> wrote: |> |> TCP was in RFC 1035 already. |> The problem is that (a) it is the fallback for resolvers, (b) the |> UDP part may already have sent two (v4, v6) packets before instead |> of only one, and (c) that PQ algorithm is simply a brain damage |> use case for DKIM, which is why DKIM does not use it. | |You misunderstand the nature of the "fallback". TCP is used when a |UDP query returns "TC=1", rather than after multiple UDP queries |time out...
I do not understand why you repeatedly bang that drum. It is exactly as i say, and it is exactly as you say. TCP for a normal stub resolver comes into play after (a) the UDP part has already sent out one or two (v4, v6) packets. In addition this happens behind the scene of the (IETF /) BSD resolver interface, there is no way to startup directly in TCP. This is how it was in about Y2K when i got there, and it is how it is today. And it is just fine that way, generically speaking. And this *could* be true with DTLS/DNSSEC even for the non-forseeable future, if today's engineers writing standards in the according parts of the IETF (luckily not the email one!, as i see it) look over the rim of their cup; most of them surely do. (And even surely more than i, and with more understanding.) So it is like i said in i think another ML (mailmaint or emailcore), to satisfy the monstrosity this WG develops (i have not yet read the last update(s)) you need to ship your own resolver. --steffen | |Der Kragenbaer, The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt) _______________________________________________ Ietf-dkim mailing list -- [email protected] To unsubscribe send an email to [email protected]
