Leo Bicknell <bicknell(_at_)ufp(_dot_)org> writes:
Ok, I'll admit I don't have all the facts, but I don't see how this
situation can exist with what has been proposed.
Here, maybe this will help:
* News servers currently always use one method to forward postings to
moderators, namely they add a To header to the message as posted and
dump it into e-mail. (There are a very small handful of exceptions,
but not enough to be worth thinking about much.)
* Most moderated groups use ASCII, but not all. But even the ones that
don't at present use the above method, which means that in hierarchies
where using unencoded 8-bit characters is common, they get them via
e-mail as described above. It seems likely that the only hierarchies
in which there are moderated groups and unencoded 8-bit data in headers
are using character sets that avoid the range of characters that get
mangled.
* Moderation software is often running nearly on autopilot, and there is
no standard package that everyone uses. A bunch of people use STUMP,
some people use DMOD, and then most moderators have rolled their own.
Many moderators just use a regular e-mail client and post things
manually, which works fine for low-traffic groups.
If news is UTF-8, raw, there must be some mechanism to e-mail that
content to a moderator,
News is not UTF-8, raw. This is just a proposal, which Charles is
advocating and which some people on USEFOR think is the best solution.
How to deal with the implications of this position for moderation is not
something that has yet been fully fleshed out, although the current theory
(as I understand it) seems to be that everything will get translated into
RFC 2047 by the news server before being sent to the moderator.
This is probably doable, but I'm skeptical.
have them be able to read it at least well enough to decide how to
moderate it, and then be able to send it back in some format that
preserves the content with their additional approval (which most likely
implies UTF-something).
It is *not* necessary for moderators to be able to read or understand all
the content in the headers to be able to moderate a group. Consider, for
example, the (likely increasingly common) case where the non-ASCII content
in the headers is restricted just to the person's name, because they don't
want to use an ASCII transcription of their real name. I don't need to be
able to read the From header to be able to moderate the group; all I have
to do is be able to get it into news without mangling it.
But that being said, the problem isn't just what to do about postings that
contain non-ASCII headers. The problem is how to allow news servers to
continue to use *one* method to forward messages to moderators rather than
making moderators handle two possibilities depending on what the message
looks like, and simultaneously not breaking the moderation of all the
existing ASCII moderated groups.
That leads to a desire to be able to continue to send messages to
moderators using the current method.
Another possible alternative would be to handle moderation differently for
groups with non-ASCII character sets, although I don't think this actually
scales well because of the above example of someone who just wants to use
their correct name.
Any solution that changes the way that posts are sent to moderators for
the existing ASCII moderated groups is going to break a *lot* of software,
and the proposal of that in the past caused massive pushback by
moderators.
IMHO moderation is broken anyway, because it must go to e-mail and back.
I doubt very many people would disagree with you that it's less than
ideal, although I'm not sure that's the right point to try to fix it.
Fixing this is *way* outside the scope of any current working group,
though, and isn't something that would happen at all quickly, even if we
could come up with a better protocol.
(It would certainly be better from a cleanliness perspective if all posts
sent to moderators were encapsulated rather than gatewayed, for example,
but this unfortunately is likely to be something that we're too late to
fix.)
--
Russ Allbery (rra(_at_)stanford(_dot_)edu)
<http://www.eyrie.org/~eagle/>