A note on how things work:
- incoming articles to ferret-list have their headers
canonicalized (in essence, strip everything except From:
Subject: and Date:), and appended onto file "article".
- At 5:02am, if there is a file called "article", a
header is computed with date, article number and current
subscription size. The header, article and trailer are
then concatenated as mail into the outgoing mailing list
alias.
- The outgoing mailing list list consists of:
- several anon users (about 10)
- several mail-to-something gateways (about 4)
- Bill's exploder at cuny, which explodes it out
to about 210 subscribers. (Thanks again Bill!)
Anon articles get dumped in the requests mailbox. Requests get dumped
into the requests mailbox. I fire up the mailer on the requests
mailbox, issueing commands to:
- append anon articles onto the current "article" file,
then edit the article file to "fix" the headers.
- issue commands to add or delete subscriptions - I've
written an incredibly braindead shell script that
prompts me for the particulars, adjusts my lists,
and sends off exploder add/delete requests as appropriate.
I've decided to keep the up-to-date list here, and update
both it and the Cuny exploder simultaneously rather than
let people subscribe/unsubscribe by themselves with
the exploder.
Making anon automatic is actually trivially easy - whacking off
the From: is no trouble. *Except* I can't be sure that people
haven't included a signature. Which is problematical to detect.
I will probably eventually automate it, but I'll have to issue
warnings to people about what possible pitfalls there are.
Bogus posts? I'd rather not talk about what the software does about
those....
Traffic? Well, we were going pretty heavy, slowly increasing,
til I went to Japan, and the roboposter broke (bad alias file
entry). When it got fixed, the load was very low for some time,
and has slowly been rising again. It'd be interesting to
chart it. Traffic is very much a function of critical mass
and positive feedback.
Nancy & Katie: Speaking as someone who's been doing this sort of stuff
for too many years, the only reliable way to build a combined (shelter)
FAQ is to have one maintainer and n contributors. Or a separate list.
Having two maintainers of one FAQ is impossible. Standard source code
control principles apply. I agree with your decision to go as
a separate file. With some care, perhaps you can make your
list distinct, and be able to exchange entries with STAR.
[Posted in FML issue 0505]
|