On Thu, 25 Oct 2001 17:16:50 PDT, Bob Fink said:
So, please let the ngtrans list now if you have any problems with
forwarding this draft, preferably by 1 November.
Technically, I see no major issues - there are a few places where I don't
think we *know* the "best current practice" yet, but those are all carefully
worded with "implementation dependent".
In section 5, there is a discussion of SERVFAIL returns from broken DNS
servers (which I've seen myself in production). That could use a one-sentence
clarification that the problem is the DNS server only having A records for
an address, and returning SERVFAIL rather than "no data" when asked for AAAA
We all know who the offender is, and those of us who use Sendmail already
know why 'WorkAroundBrokenAAAA' exists. I'll let somebody with more RFC-writing
experience than I decide if package-specific "If you run the XYZ MTA, do THIS
to work around it, if you run the ABC DNS server, upgrade to patch 1.15" info
should be included in an "operational requirements" RFC. RFC1912 mentions
specific releases of BIND, RFC2010 doesn't.
Although Motonori's English is vastly superior to my Japanese, I still saw
a few places where phrasing was awkward, or verbs had case/tense issues. I'd
consider them not a show-stopper, but if a volunteer from the audience were
to go through and clean them up, the document would benefit (I'll be happy
to do so if nobody else wants to, but consensus calls for a clean-up - but if
nobody else cares, I'll let it slide).
Operating Systems Analyst
Description: PGP signature