-- On Monday, April 14, 2008 8:58 PM +0200 Frank Ellermann
<nobody(_at_)xyzzy(_dot_)claranet(_dot_)de> wrote regarding Re: IESG Statement
Spam Control on IETF Mailing Lists --
Russ Housley wrote:
When IETF lists are housed somewhere other than ietf.org,
they are supposed to include an archive recipient so that
there is an archive available at ietf.org
Makes sense. I have submitted some lists to "other lists",
how is this archive recipient magic arranged ?
I can tell you what is supposed to happen. The short answer is
that RFC2418 tells you one of the two email addresses to subscribe
to your "other list" to get an archive on the IETF web site.
However, I think that guidance is at best incomplete. The rest of
this message is the long answer and why I think that guidance is
RFC2418, IETF Working Group Guidelines and Procedures, says this
in Section 2.2 Charter:
As a service to the community, the IETF Secretariat operates a
mailing list archive for working group mailing lists. In order
to take advantage of this service, working group mailing lists
MUST include the address "wg_acronym-archive(_at_)lists(_dot_)ietf(_dot_)org"
(where "wg_acronym" is the working group acronym) in the
mailing list in order that a copy of all mailing list messages
be recorded in the Secretariat's archive. Those archives are
located at ftp://ftp.ietf.org/ietf-mail-archive. For
robustness, WGs SHOULD maintain an additional archive separate
from that maintained by the Secretariat.
In turns out that this guidance is both incomplete in practice and,
unfortunately, wrong in one detail.
The incorrect detail is that the IETF no longer uses the domain
lists.ietf.org. In fact, it never used it correctly. During the
transition we discovered 7 domains used for mailing lists, the
predominant one being the domain "ietf.org". As part of the
cut-over it was decided to use the domain "ietf.org" exclusively
for all IETF lists. Everything was setup that way with backwards
compatibility in place for all existing lists that used some other
domain. Today, all lists are created in "ietf.org", unless they
are IAB, IESG, or IRTF lists.
In practice, there are a few incomplete details.
First, there are actually two aliases that need to be subscribed:
Some might ask how this separation came to be? I don't know and
never asked. It is continued today mostly for convenience; it's
quite tedious to "undo" the current setup and there are more
important things to be done.
Second, except for the guideline in 2418, there's no process or
documentation to go with this whole model on the IT side. Here's a
sampling of some questions.
1. There's no maintenance of this practice. Last year the IETF did
an audit, for the first time in forever, and brought everything
up-to-date. This means that at that time all the "other lists" had
an archive on the IETF web site. However, we're out of date again.
2. These archives have a SPAM problem. Well, they had a much more
serious SPAM problem. Early in the post cut-over process AMS (Glen
Barney) added some SPAM protection to these archives. However, all
the best work has been put into getting Mailman firmly entrenched
and protected. My opinion is that the IETF should just create a
mailing list for every WG and then these "other lists" should just
subscribe the IETF list to their list. This way there's no "extra"
work on the IT side to protect things, particularly since Mailman
provides some useful built-in features. In addition, we don't need
these "magic" aliases any more.
3. Part of the maintenance problem is that for obscure reasons the
IETF subscriber on these "other lists" will drop off the list
occasionally. I'd be interested in hearing any good ideas for how
to deal with this issue.
4. The Secretariat has to take action to make these email addresses
work. I realize this is probably obvious to everyone here, but in
the interests of completeness it should be documented that if
you're going to setup an "other list" you need to ask to have the
archives created. Or, perhaps, a back-office process should exist
(does not currently) that says these archive should be
automatically created whenever a working group is created.
5. These questions and issues are frequently asked or framed in the
context of WGs. However, on the IT side, they apply more broadly,
i.e., they apply to BOFs, directorate mailing lists, and other
private lists. This not normally visible to the
community-at-large, and perhaps should be at least accessible in
some way even if it's not announced, but my comment here is that
none of it is documented in any way.
IETF mailing list