ietf-822
[Top] [All Lists]

Re: UTF-8 over RFC 2047

2003-01-15 08:55:58

Russ Allbery wrote:
Charles Lindsey <chl(_at_)clw(_dot_)cs(_dot_)man(_dot_)ac(_dot_)uk> writes:


Nobody here has been prepared to say that they expect RFC 2047 encoding
to be around in 20 years time.


I expect to see RFC 2047 encoding still in use in 20 years time.  I
believe I already said that.

I definitely said so. In fact in 
<3E171972(_dot_)2060801(_at_)alex(_dot_)blilly(_dot_)com> I wrote:
   we will continue to need mechanisms like 2047 / 2231 [yes, Charles, even
   20 or 30 years hence

My position is basically this:  Usenet does not have a lot of mind share
or a lot of market share, but it continues to be very useful for some
particular areas.

Agreed.

There are significant problems
on Usenet that need resolution, such as [...]
>  improving the linkage between
mail and news for gatewaying, and the whole specter of authentication and
authorized article deletion.

Indeed, *improving* the compatibility of mail and news (rather than incompatible
divergence) ought to be addressed.  And implications of any proposed changes on
authentication should be given adequate consideration now -- any proposal that
requires changing message content after signing (e.g. encoding in gateways) will
simply not work, and will only cause severe problems when authentication is 
fully
addressed.  Just-send-8-bits-and-fix-it-in-the-gateway may make life easy for
lazy UA programmers now (and make life impossible for gateway developers), but
it will result in severe problems.  Note that the Usefor charter lists signed
articles as the first of the "areas [which] need urgent attention".

Time spent trying to solve problems that are *not* unique to Usenet and
that don't further any of the things that make Usenet interesting and
useful as a protocol suite are not nearly as interesting.  In fact,
they're a huge distraction and waste of time and effort for the few people
who are developing *for Usenet* to have to re-solve problems already being
addressed by other protocols.  This is *particularly* true when it comes
to problems that are already being solved with e-mail, given that one
reason why there are as many news clients as there are is because it's
fairly trivial to write a news client once you've already written a decent
mail client.

From my perspective, solving problems that are related to the Internet text
message format as extended by MIME and other RFCs is interesting.  Under the
current news article format standard (RFC 1036) that includes news (there are
a few relatively minor special cases -- and one goal of Usefor ought to be
a migration toward *better* compatibility), but there are no insurmountable
problems.  If there is eventually an article format which deviates
sufficiently from the Internet text message format, I have no plan to
adress that format, precisely because Usenet is such a small part of the
"big picture".  Show-stoppers include subtantially-different definitions
for a common header field (e.g. parameters, raw 8-bit content in message
and MIME-part header fields), redefinition of existing MIME media types
such as message/rfc-822; which are in the current Usefor draft.

[some material elided for brevity -- no disagreement]

In other words, all of these debates are very important, but I think that
so far as possible, Usenet should simply say "this is a different way of
transferring e-mail messages with a few additional fields" and be done
with it, just like RFC 1036 did.  People who feel strongly about
particular ways of handling such common problems as internationalization
should contribute actively to the e-mail standards world and Usenet will
*automatically* follow further developments in e-mail.

Agreed wholeheartedly.

As an implementor, I would dearly like it if people involved in Usenet
standardization work would stop trying to get me to work on problems that
are not *Usenet* problems and instead let me use the mass of existing
techniques, libraries, and tools for handling the e-mail message format
and concentrate my work on those parts of Usenet that actually make it
*Usenet*.  (Like newsgroup names.)

As an implementor of libraries and tools for handling the Internet text
message format (email, and currently -- under 1036 -- news) I would dearly
like it if people involved in Usenet standardization would stop trying to
deviate from that common format. I'm happy to let specialists in respective
areas deal with display issues, GUIs, voice messaging esoterica, and specifics
such as those related to Usenet (e.g. newsgroups and distributions)



<Prev in Thread] Current Thread [Next in Thread>