[emv] Administrivia: no repostings for a while, comments ?
Edward Vielmetti
There will be no more repostings of articles into comp.archives until
some time in early February. If you have a package that you would
like to announce, and you have (successfully) relied on comp.archives
to announce it here for you, then it might be good to send the article
off to this group yourself. I'm off for a week for Usenix, then a
week's vacation, and it's not clear how much time I'll have
immediately following for reposting. (That plus there's a two-week
expire running here at math, so some stuff might fall off the edge.)
I imagine that people find comp.archives interesting and useful; it
seems that readership is pretty high, and the comments I've received
so far have been generally positive. My repostings are still
"experimental", i.e. there's room for improvement. Let me describe
the current technique in brief.
My news reader is "gnus", the emacs nntp news interface. It's pretty
good. In particular, it has KILL files that can effectively look
for only articles which match a certain pattern, something that
proved quite difficult in 'rn'. Here's a sample:
(gnus-kill "" "FTP\\|^%A" "u")
(gnus-kill "Subject" ".")
(gnus-expunge "X")
This is from comp.ai.KILL; it's pretty basic, just finds all of the
articles with FTP or %A (refer format) in them, marks them unread,
marks all the rest as read. Do this for every "interesting" group
that you don't have time to read all of and voila, an effective
filter. Without this technique or something similar it would be
unfeasible to do what I'm doing.
Next is to feed the interesting subset of these articles into
news. The technique for this is "min", a mid-sized perl program
which rewrites headers and tacks on other headers as I see fit.
I invoke it as follows on a typical article:
min -g arch -n min.perl/not-available-yet \
-s "min, repost articles to moderated groups" \
-f derrida.math.lsa.umich.edu -c pub/nil \
-m "min.shar.Z"
(actually it's usually somewhat simpler than this but this is
the basic idea). It's not ready for public use yet, though if
you're moderating a group or compiling a digest I'd like to
share it with you.
Finally the problem is to extract the information from several
months of postings like this and put it into an easily browsable
database. I have ideas on this one, but for the time being
the best tool is "grep". The "Archive-name" tags are intended
to be meaningful and unique, but (sigh) not every posting has
one. Any contributions to the art of archiving are appreciated.
--Ed
Edward Vielmetti, U of Michigan