--On Tuesday, December 24, 2002 11:10 AM -0800 Russ Allbery
<rra(_at_)stanford(_dot_)edu> wrote:
In any event, there is widespread feeling on the USEFOR list that failure
to provide for Unicode newsgroup names would basically make the entire
effort of producing a new standard pointless, and all of the solutions
seem to have significant problems. There seems to be a consensus that
using raw UTF-8 in the Newsgroups header for Usenet articles is the least
bad choice.
I think the current text on gatewaying such messages to mail is completely
insufficient. The algorithm given in 5.5.2 won't display in any clients
without modifications, despite those clients having UTF-8 support.
My gateway is an IMAP server which is used by clients to read netnews. IMAP
is a superior protocol to NNTP for end client access (it has server-side
seen state, secure authentication, and other features) and it would be nice
not to lock this out.
The majority of e-mail clients interpret untagged 8-bit in their local
character set, which is likely not UTF-8, so ignoring the standards is not
a reasonable or acceptable answer. Using message/rfc822 to identify
articles (as in 6.21.2.2) with unencoded 8-bit headers is highly harmful
for processing messages.
Section 6.21.3 says
Injecting and relaying agents MUST NOT change the encoding of
articles passed to them. Gateways SHOULD NOT change the encoding
unless absolutely necessary.
Please reconcile this with RFC 2822; you are already identifying non-RFC
822 messages are "message/rfc822" _and_ you say that gateways shouldn't
change this. What about clients saving messages to a local filesystem?
Should the mime type be identified as "message/rfc822"?
Larry