On Wed July 13 2005 05:10, John C Klensin wrote:
--On Tuesday, July 12, 2005 21:22 -0400 Bruce Lilly
Some quick comments on the draft and its security counterpart.
1. numeric reference tags are deprecated, as discussed on the
rfc-interest list several months ago
One-line change in the XML, which has been tested. I decided to
leave -00 as is to permit more easy comparisons to 2821.
Given that the reference tag numbers differ and the citation text
also differs, I fail to see how that facilitates comparison.
And discussion on rfc-interest is not
exactly the way normative changes are made around here.
Changing the format of a reference tag is hardly normative; it
is at most editorial. The comment was simply a reminder (you
were aware of the discussion, as you participated mid-January).
2. the security draft needs a good proofreading:
O "among" should probably be "amount"
O " if IDENT" should probably be " of IDENT"
Which part of it did you have trouble understanding?
On first reading there was something that made no sense; I could
not find it again during a quick scan when composing my comments.
I am leaving on business travel in a few hours, so I don't have
time now to look thoroughly.
differently, it was a hasty job to aid discussion and the choice
was between getting it out as is or doing the careful
proofreading and posting in around 8 August.
A spelling check would probably have caught many of the typos,
should have taken at most a few minutes, and would save many
reviewers much time trying to decipher the text.
3. There remains a conflict with RFC 2822 w.r.t. the Received
field; 2821[bis] permits a quoted-string for the "id"
component, which is not permissible in the field as defined
in RFC 2822.
Clearly one of them should be adjusted. Which one is, IMO, a
list or WG discussion item.
If the fundamental concept of 282, i.e. consolidation of
documents rather than introduction of new features, has not
changed, then the course of action is clear; quoted strings
were not permitted in ID by RFCs 821 or 822. Indeed, a
number of constructs that were legal in 822 in various places
have been deprecated to simplify parsing; I know of only one
case where a new feature was added, and that was a concession
to widespread (but illegal) use -- I know of no widespread
use of quoted-strings in the ID part of Received fields.
And I would remind you that this particular discrepancy was
noted shortly after publication of 2821 and 2822 four years
4. The same peculiar and confusing wording of RFC 1123 is
repeated; RFC 822 did not "suggest" an '@' in a msg-id, it
clearly specified it (along with '<', '>', '.', and
possibly '[' and ']') in msg-id.
And this is not 822 or 822bis.
The comment isn't about 822, it's about text in the draft
which mischaracterizes 822.
5. "TCP" really ought to be registered with IANA for use in the
"via" Received field component (or dropped, since it wasn't
Yes, someone should do that. Do you think it can be just done,
or that we need to create a new "IANA considerations"
One issue is whether it should have been introduced at all, since
it had never been registered and was not in 821 or any of the
documents which amended 821.
It was introduced in 2821, but not registered.
Your draft states:
"Via" is primarily of value
; with non-Internet transports. SMTP servers
; SHOULD NOT use unregistered names.
So as it stands, "TCP" SHOULD NOT be used, since it is
*If* it is to be introduced as a new feature, then it ought to
be registered, and that takes one sentence (not subsection) in
the IANA considerations section.