I'm posting this to comp.lang.lisp only. If someone who feels part of
the comp.lang.ml community wants to post a similar item there, that's
up to them. That's not my community, so I'm not going to opt to
intrude there.
I just want to note as an aside that I (and perhaps others) as a
matter of principle do not respond materially (i.e., "on topic") to
any conversations that are cross-posted. As such, any data that you
think you're getting about the opinion of the Lisp community by
cross-posting to the Lisp community is biased by the fact that you
have elected to talk only to that subset of the Lisp community who is
comfortable cross-posting--not a scientific sample.
I believe it is sometimes appropriate to cross-post an advertisement
to such a broad distribution--saying "come to the such-and-so discussion
on some newsgroup"), but follow-up should be to just that group.
I find the various newsreading tools I know to be hopelessly bad at
tracking such distributed conversations in a coherent way and I thus
prefer not to add to the chaos myself. I end up with multiple
killfiles trying to silence a certain conversation and its offspring
selectively for some newsgroups but not for all because I want to
track it on one, but as it spawns children, they again recur on all
lists and it gets messier and messier. In sum, I think cross-posting
creates a sufficient technological burden on a sufficient number of
people so as to be regarded as antisocial on its mere face.
But further, I also think that in general there is a resource mismatch
because anything cross-posted is RECEIVED BY the UNION of people
interested in both groups but probably OF INTEREST TO only the
INTERSECTION of the two groups. I think this places an undue burden
on the members of both groups to tolerate out-of-band discussions that
often they are simply not interested in. And as a matter of
principle, I decline to participate in such a forum because of its use
of resources unbefitting the probable importance of the conversation
to the group.
I feel that as the audience size increases, the amount of common
ground that can be presumed decreases. I feel comfortable talking to
Lispers because I know what I can presume in terms of their
background. I am capable of talking to non-Lispers but it affects the
data transmission rate. Talking to a mixed group just makes some
things I say hard on one community because they're too complex or hard
on the other because they're too paintstakingly described. Overall, I
think that suits no one.
And, finally, I feel discussions don't tend to terminate when they
happen across-forums because what it takes to satisfy one forum is
often different than another. These are, in a way, not unlike
political parties and although we like to all be friends and allow
each other into our forums, I think the unspoken rule is "do not come
to my forum if you are not interested in my topic". So when someone
appears on the lisp forum and says "why do you guys always talk about
lisp" I get bent out of shape--I feel I have a right to just talk
among people who don't make me justify my NEED to talk about Lisp and
who are simply there to talk about HOW and WHY a particular Lisp
solution is good. WHen a conversation becomes cross-posted, one must
not only justify the thread topic, but ultimately justify the slant
that the community of the poster would take for granted. And often I
think the answer is "agree to disagree" which is what separation of
forums implement. By putting the forums together, that natural
"agreement to disagree" is undone and one gets more of a "perpetuation
of disagreement".
I always get the impression that people thinking they are being
"friendly and inclusive" by having a discussion across multiple
forums who they are oh-so-sure are interested. But, in fact I find
such discussions EXclusive because the price of participating is to
feel you are rudely intruding on too many others.
Indeed, I do not even READ cross-posted messages because I find them
rude. I wish all cross-posted conversations would go away. If anyone
has the Emacs lisp code to make cross-posted news postings get killed,
I'd like to see that. I tried (gnus-kill "Newsgroups" ",") but it
complains that "Newsgroups" is not a header field.
I would DEFINITELY have responded in substance and on-topic to the
Static Typing discussion if it were a comp.lang.lisp-only discussion.
But because it was cross-posted, I feel excluded.
From: Erik Naggum
Subject: Re: The cross-posted Static Typing discussion
Date:
Message-ID: <3089536619600484@naggum.no>
* Kent M. Pitman
| I wish all cross-posted conversations would go away. If anyone has the
| Emacs lisp code to make cross-posted news postings get killed, I'd like
| to see that. I tried (gnus-kill "Newsgroups" ",") but it complains that
| "Newsgroups" is not a header field.
that looks like a pretty old Gnus, but maybe this applies to old Gnus
versions, too: Gnus reads headers with the NNTP command XOVER, which
returns entries (lines) from a database of (tab-delimited) extracted header
values called NOV (News OverView). which headers are extracted is a system
configuration option, but certain headers are required and standard. I
believe Xref is among those headers, provided that your NNTP server adds it
to incoming articles in the first place. (it's a local header, like Path.)
the Xref header looks like this (your article, at my news server):
Xref: ifi.uio.no comp.lang.lisp:29614
with additional newsgroups and article numbers appended in like manner.
it follows that a cross-posted articles differs from a non-cross-posted
article in that it has at least three colons in this header line. (two
colons if we only look only at the header's value.)
so (gnus-kill "xref" ":[^:]+:[^:]+:") should kill cross-posted articles,
leaving non-cross-posted articles behind. it works like that here, at
least. you might want to check that articles actually have sane Xref
headers and that the NOV database has Xref headers.
#\Erik
--
if you think this year is "97", _you_ are not "year 2000 compliant".
see http://sourcery.naggum.no/emacs/ for GNU Emacs 20-related material.
From: Thomas A. Russ
Subject: Re: The cross-posted Static Typing discussion
Date:
Message-ID: <ymik9dvtjjk.fsf@sevak.isi.edu>
I use a slightly different method than Eric. Instead of looking for a
header that may or may not be in the configuration, I chose less
efficient filtering query that scans the entire text of a message. This
forces the newsreader to download and read all the messages, but for
some newsgroups (like the lisp one) this is usually not too much of a
burden.
(gnus-kill "" "^Newsgroups:.*,.*,.*,")
The proper number of commas can be inserted. Unfortunately, there are
some people who post to both the .clos and .lisp groups, so you may end
up with too small a set of messages if you set the tolerance too tight.
--
Thomas A. Russ, USC/Information Sciences Institute ···@isi.edu
From: Richard Hoskins
Subject: Re: The cross-posted Static Typing discussion
Date:
Message-ID: <m3u3cz5cxr.fsf@interlaced.net>
···@sevak.isi.edu (Thomas A. Russ) writes:
> I use a slightly different method than Eric. Instead of looking for
> a header that may or may not be in the configuration, I chose less
> efficient filtering query that scans the entire text of a message.
If you set `gnus-nov-is-evil' to non-nil, Gnus can score articles
based on the xref header even if they are not included in the overview
database.