From: Bruce Lilly (blilly@erols.com)
Date: Wed May 21 2003 - 14:32:43 CDT
Pete Resnick wrote:
> Let me start off by noting that Bruce is the only one arguing for
> removal of this restriction
To be precise, I'm not arguing for removal of the restriction, which
would be either removing the text completely or changing the MUST to
MAY. As I have stated earlier, relaxing the restriction, i.e. changing
MUST to SHOULD, would be adequate to resolve the conflict w/ RFC 2047.
> the barrage. (On this topic alone in the past 5 days, Bruce has sent 7
> messages , Charles has sent 4, I have sent 4 [though two of them were
> only 1-2 sentence queries], and Russ and Bill have sent 1 each.)
Point noted, however, many of my messages were responses to queries,
or clarification necessary in the face of misinterpretation or
misrepresentation, or responses to queries on side issues (e.g. how
SMTP deals with messages that don't end in CRLF).
> All: Please try to allow other people to participate and hold your
> responses for a while to see if anyone else is supporting you or whether
> you're completely off base. The current discussion is not doing any
> persuading nor is it trying to reach consensus.
Certainly I welcome other comments. But the root issue of the conflict
with 2047 needs to be adequately addressed if the restriction is to
remain a MUST. Inserting a comment is a nice ivory tower daydream,
but I am unaware of any posting agent that does so, nor of any existing
gateway which does so.
> It's not just every field; it's even quite a bit of the field contents
> (i.e., some of the lower-level ABNF) that has to change. That would be a
> bummer.
No argument there -- I've said that the incompatibilities require a
substantial amount of work to document. That's why I prefer to minimize
the incompatibilities.
> A comment before the mailbox is not one of the permitted forms in 1036
> *to specify a full name*. There is nothing in the text that forbids it,
> just as there is nothing in the text that forbids multiple whitespace
> before the comment even though it's not shown in the permitted forms.
> Agreed that it will confuse parsers who are looking for a full name, but
> that's mostly a display issue.
It is much more than a display issue The From field isn't there to look nice.
It may be converted into a body text attribution line (which may look
silly if it's botched, but is otherwise harmless). More importantly, it will
be converted into a recipient field (To or Cc) if a reader decides to email
a response. There are things permitted in a comment which are incompatible
with such a field. If the comment text becomes part of the recipient address,
the message may be undeliverable.
All of which is moot because, as noted above, no known existing posting agent
or gateway inserts a comment in such a case.
> Let's also agree that there is a significant difference between breaking
> proper display of a full name and non-delivery of a message.
Yes, agreed, but as noted, we're not dealing with mere display. A message
that doesn't propagate on Usenet via INN may well find another route via
a different NNTP implementation, or via UUCP or some other transport, due
to the flooding means of propagation. A botched recipient address for an
email response simply won't work at all. So in that respect inserting a
comment (even if that were being done) might do more harm than good.