On 2026-01-18 at 12:51 -0500, John Levine wrote: > It appears that Joe Abley <[email protected]> said: > > > However, I am in favor of adding a text such as the following: > > > > > > [...] > > > > I quite like your formulation, and I think I agree that it's better > > than what we suggested initially. Thanks for > > taking the time to write that up. I think based on other comments > > consensus was already tilted towards making > > more tightly-scoped advice around CNAME and DNAME and being more > > explicit about not changing the packaging of > > DNSSEC RRs, and your text does this quite elegantly. > > > > I am interested to hear whether anybody else thinks differently. > > I think it's fine, but I also still think we should tell the stubs to > grow up and deal with > the records in whatever order they arrive.
If we are simplifying the protocol this way then there is no point in them doing that, which requires added complexity and they have lived perfectly without supporting that so far. In fact, I wonder how many other bugs are lurking in resolvers supposedly supporting CNAMES in any order. > I realize that neither the caches nor the stubs will all be fixed any > time soon, but a stub that you can crash by sending it the right > answer in an unexpected order is a pretty bad piece of software. I completely agree with this, though. Crashing is not acceptable at all. _______________________________________________ DNSOP mailing list -- [email protected] To unsubscribe send an email to [email protected]
