On 2022/02/17 11:16, Dave Voutila wrote: > > Evan Silberman <[email protected]> writes: > > > OpenBSD is distributed with an authoritative name server daemon, nsd(8), > > but that distribution doesn't document the format of zone files in > > manual section 5. I thought hey, you should be able to read about zone > > files in the manual. So I took a stab at writing zonefile.5, attached. > > > > I did my best to address nearly all important details of zonefiles as > > standardized, as well as a useful subset of the RR type zoo. Included is > > a description of SRV records, which might not be useful/relevant enough > > for most administrators. Excluded is a description of CAA records, which > > are probably increasingly useful to cautious administrators but which I > > haven't used and seem to have more complexity than I want to boil down > > for a manual. > > > > I am interested in feedback in general, and also specifically on the > > following: > > > > - have I included more detail than OpenBSD might want to be responsible > > for documenting? > > Have you offered similar documentation to the upstream nsd project from > NLnet Labs? https://github.com/NLnetLabs/nsd > > > - have I excluded any RR types that many people will want to read about? > > - have I used mdoc(7) properly? (I use Ic to mark up RR types and Ar to > > mark up the fields of RDATA. Makes sense to me and renders nicely but > > I can understand wanting to dispute this idea.) > > - have I described an RR type or some clarification that's in an RFC I > > didn't reference? > > > > I'm happy to respond to feedback and resubmit complete drafts, but if > > this is promising enough for OpenBSD developers to commit, I'm happy to > > send diffs in response to feedback as well. > > > > I don't work on nsd(8), but getting this into the upstream project and > maintained there would probably make the most sense. I will admit that > since the file isn't inlined in the email, I didn't inspect it so won't > speak to its quality at the moment.
They use man formatting not mdoc unfortunately. Could be converted but it's not as *nice* :) (It's equally relevant to BIND and Knot, and at least partially to PowerDNS, all of which seem to use RST as their "base" format for docs). > > If nothing else, this was fun to work on. I've only had a quick read but I like it so far. At least NSD, BIND, knot currently defer to the RFCs for this (mainly 1035 but then you have to pick through a couple of dozen 'updated by' to figure out which extra parts apply) and they're not so informative about real world use, this seems a nice place to mention "gotchas".
