Thanks Guozhen, I have put comments below... > On 23 Sep 2026, at 04:06, Guozhen Dong <[email protected]> wrote: > > I've read draft-mk-svcb-well-known-00. A clean, well-scoped draft. Defining > "well-known" as a general primitive that other specifications can reference > is the right design, and the motivating cases are persuasive. > > I have a few questions: > > 1. Section 2.2, order preservation is specified, but not whether the same > suffix may appear more than once in a list. For example, "A suffix SHOULD NOT > appear more than once in a list; a decoder MUST tolerate duplicates"? Or is > there a reason to leave this unspecified?
There is no reason to leave this unspecified. A well-known suffix would only appear once, so we can tighten up the draft here, thanks :-) > > 2. Section 2.2, "A suffix may extend below a registered well-known name where > that registration allows it". How does an implementer tell whether a given > registration allows subpaths? Would a concrete example be helpful? The implementor would need to check the IANA registration to determine this. For instance the posh registration might allow "/.well-known/posh/spice.json” so the SVCB well-known parameter would be: well-known=posh/spice.json We can add this as an example. > > 3. Section 5, the draft advises publishers to keep advertisements compact. > Would a brief operational note be added, or is the number of suffixes best > left entirely to the publisher? We can look and see if a note is needed, it would suggest that we perhaps say something along the lines of values for well-known MUST be available at the service, again we’ll take a look at this section. > > 4. Section 7, I'd like to register support for the suffix-only choice; it > matches the form used by IANA "Well-Known URIs" registry and saves octets on > the wire. No change requested—just noting a preference among the open issues. Thanks for that, we think it is the best choice. Appreciated the feedback, Jim _______________________________________________ DNSOP mailing list -- [email protected] To unsubscribe send an email to [email protected]
