On Mon, 1 Jan 2001, Ian Clarke wrote:
> On Mon, Jan 01, 2001 at 06:52:17PM -0600, Mark J. Roberts wrote:
> > I do! I do! I think we should all eat our own shit^H^H^H^Hdogfood and
> > distribute the mailing list with Freenet! (HTML archives, too.)
> >
> > If I don't have a working and easily configured (read point at list URI
> > and it configures itself) daemon by 12:00AM the day after tomorrow, you
> > have permission to slowly, painfully kill me. I've already got a working
> > insertion daemon, more or less. Mail spool parsing is a bitch. A line like
>
> Does it intelligently search for the latest message using that algorithm
> I suggested ages ago? (ie. trying 2, 4, 6, 8, 16, 32 - oops, nothing at
> 32, 24, 20 - oops, something at 20, 22 - oops, nothing at 20, 21 - oops,
> something at 21 - 22 is the lowest unused slot).
>
> Ian.
I use MSKs inserted under our new automatic date-updating SSK redirects.
The daemon watches a mail spool file and inserts any new messages at a
configured interval (it can do it in advance of the date under which it's
inserted, so the insert has enough time to get into Freenet).
I insert the raw messages (filtered and cleaned up a bit) under the
filename "raw" in the mapfile. The raw messages are inserted every time.
The html archive, which contains the past n (configurable) messages, can
be inserted under "index.html" every time or every n (configurable) times.
If a new index.html is not inserted, a reference to the old one is. This
might be useful because you might want to insert the latest messages in
raw format (for subscribers) every 10 minutes, but you only want to insert
the bigger index.html archive file every hour. How often to insert each is
controversial -- you don't want to wait too long to recieve the latest
messages, but you don't want to have to constantly make requests,
especially with Fred (at least without gcj/libgcj). Then again, your mail
client requests new messages every minute or so. But only from one place,
your ISP's mail server, while a subscriber to five Freenet lists each
updated every 10 minutes needs to make a request every 2 minutes.
Shouldn't be a big deal with reasonably efficient Freenet clients. I think
10 minutes is reasonable. 60 minutes for web archives. But that's all IMHO.
This way we get really kewl archive URIs like
"freenet:lists/freenet-dev//" that can be accessed in web archive format
with your browser or subscribed to with a simple command-line option
"-subscribe freenet:lists/freenet-dev//". Very simple and intuitive.
If I'm lucky or especially motivated we'll also have automatic listings of
past archives (albeit slightly overlapping message content) but that'll
require cramming the addresses of the previous archives in a data file
for storage. Probably won't have time for that one in the next two days,
but who knows.
--
Mark Roberts
[EMAIL PROTECTED]
_______________________________________________
Freenet-dev mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/mailman/listinfo/freenet-dev
