[Top] [All Lists]

RE: domain name definition in RFC2821

2006-04-22 10:39:35

Ack.  Unless others disagree with Ned's (IMO extremely coherent)
analysis, I'll pull the changes out of 2821bis.


--On Saturday, 22 April, 2006 09:37 -0700 Ned Freed
<ned(_dot_)freed(_at_)mrochek(_dot_)com> wrote:

--On Saturday, 22 April, 2006 01:27 -0700 Yuri Inglikov
<Yuri(_dot_)Inglikov(_at_)microsoft(_dot_)com> wrote:

By the way the "ws." syntax appears "invalid" according to
every RFC I read.

Except the base DNS specs and RFC 1123, which makes accepting
it mandatory everywhere.

Um, actually, not only does it not say that, it says exactly
the opposite. Section 5.2.18 describes "common address
formatting ERRORS" (emphasis mine). The fourth bullet item is:

         o    Some systems over-qualify domain names by adding
a               trailing dot to some or all domain names in
addresses or               message-ids.  This violates RFC-822

So you can add RFC 1123 to the list of email-related RFCs that
describe trailing dots as a syntax error.

I can find no reference in RFC 1123 that says accepting a
trailing dot is mandatory anywhere. The closest thing to this
I can find is the statement in section that:

           If an abbreviation method is provided, then:

           (a)  There MUST be some convention for denoting
that a name                  is already complete, so that the
abbreviation method(s)                  are suppressed.  A
trailing dot is the usual method.

But this section is talking about user interfaces, not on-wire
protocols. And it goes on to say that abbreviations need to be
expanded - and the exmaple is "a mail program". Couple this
with the previous assertion that trailing dots are syntax
errors in email and the clear inference is that the user
interface should remove the trailing dot if one is present.


That is, indeed, an issue.  But, again, 1123 is crystal-clear
on the subject.  And the 2821bis  working draft is now
consistent with it.

Well, I agree that RFC 1123 is quite clear. But I think it
pretty much says the opposite of this.

Yuri, since you seem to be looking for the bottom line here, I
suggest it is:

(1) Anywhere that "Domain" is permitted as a piece of syntax,
"Domain." is permitted.   Not supporting it is bad news.  If
you are working on this at Microsoft, note that IE supports
the trailing dot in domain names, so failure to support it in
email would presumably violate the law of least astonishment,
provide a surprising user experience, etc.

I have no problem with supporting a trailing dot in a user
interface. Indeed, I think RFC 1123 can be read as suggesting
(although not requiring) that this work.

However, I have a big problem with extending the on-wire
syntax to include trailing dots. This is a clear break from
past usage and I do not think the benefits are anywhere near
commensurate with the costs.