ietf-nntp draft-ietf-nntpext-base-17

[Russ Allbery]

Charles Lindsey <chl at clw.cs.man.ac.uk> writes:
> Russ Allbery <rra at stanford.edu> writes:

>> I don't think this is much of an issue, since the previous NNTP
>> standard mandated ASCII parameters, and the change to UTF-8 is
>> specifically noted as an incompatibility with RFC 977.  Since wildmats
>> are only applied to newsgroup names (and more than that, only to names
>> of newsgroups carried on that server), servers can just avoid UTF-8
>> newsgroup names until the wildmat implementation has been upgraded

> That's a good pragmatic approach. But you would not get away with it if
> Bruce Lilly was reading this list :-( .

I don't think he'd have a problem with it either, actually.  The previous
standard does not allow any 8-bit characters in group names since it
requires that all parameters be in US-ASCII.  There are no 8-bit group
names in widespread use at present, and those that are in use are
primarily internal (where Internet standards don't need to apply as
strongly).

The only way that one runs into trouble under this provision is if one
adopts one new portion of the standard (UTF-8 parameters) without adopting
another portion (UTF-8-aware ?).  I don't think anyone expects standards
to try to protect against people adopting bits and pieces of them in the
wrong order.

Bruce's problems are more with introducing UTF-8-encoded group names into
the normal news flow where software that has not been upgraded may
encounter them.  This may well be a real problem, but it's not an NNTP
problem, since NNTP does not address the contents of the articles
themselves.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>