archive@unmvax.cs.unm.edu (Rec.Hunting Archives) (03/28/91)
[Note: this is NOT the text for a monthly posting - that will follow later]
Introduction to the Rec.Hunting Archives
REQUEST@THEREV.UUCP
3/25/91 in the wee hours when I should be doing taxes
or sleeping, instead of tinkering at the computer
The official Rec.Hunting archives are being maintained long ago, in a
galaxy far, far away. TheRev.UUCP is a small, privately owned UseNet node
in Los Alamos, New Mexico. In Los Alamos, one may find a) the largest
single collection of Cray Research mainframes housed under a single roof,
b) this dinky generic 80286 based machine that I use, and c) a 1.43 gigawatt
generator. The advantage to the first is the incredible speed and
computing power that a researcher can bring to bear on a scientific
problem. The advantage to the second is that there is no boss or company
that tells me that I've exceeded my disk or resource quota. There is, as
yet, no advantage to the third - but I had a great time watching them set
it on it's foundation.
ARCHIVE NAMING
The archives are indexed and grouped by posting date, and accessed via
a mail-server, REQUEST@THEREV.UUCP. The archive name format is ddMMMyy -
for example, any articles with a posting date of March 13, 1991 would be
archived as 13Mar91. At the end of each month, all articles posted for
that month will be bundled into a monthly archive. I maintain an index for
each archive, as well as a master index that lists all the articles and
their archive names.
MAIL-SERVER COMMANDS
The archives are retrieved by commands IN THE BODY of an e-mail message
sent to REQUEST@THEREV.UUCP. You may put anything or nothing in the
Subject line of the mail; the server will pay no attention to it. The
server currently recognizes two commands, the send command and the special
command. Here's a list of examples for the send command:
Command What is [supposed to be] sent
------- -----------------------------
send help This illuminating and informative document
send charter The Rec.Hunting NewsGroup Charter
send index The master index of ALL rec.hunting articles,
and the archive name where they can be found.
send list The list of what's available here.
send <archive> A specific archive file
send <archive>.idx The index for a specific archive file
send <article> A specific article in the archives
send <message_id> The article referenced by the message ID
send money Has no effect - same as with parents
Why all the variations? Because I'm a hacker! The complexity in the
archives is in the indices, not in the archives themselves. The format of
the indices allows me to search for and extract just about anything without
much difficulty - the mail server is written in Borland's Turbo Pascal 5.5,
so it's simply a question of indexing the information efficiently. See
_HOW IT WORKS FAIRLY SIMPLY_ for a more detailed and probably boring
explanation.
The 'special' command tells the server to forward your request to my
personal account for special handling - it means you want me to break my
back extracting strange information for you without charging you anything
for all the time it's going to take. So use it advisedly. Here's an
special - Send me all articles on steel shot.
All that really matters to the server is that the first word on the
line is a recognized command. If you're going to make special requests, do
it in a seperate request. As soon as the server sees the 'special'
command, it forwards the mail to me. If any other commands were in the
mail, they will NOT be processed.
Now if that's not complicated enough, I'm sure I can come up with
something more obtuse. It actually works fairly simply.
HOW IT WORKS FAIRLY SIMPLY
As articles come in, they are scanned and indexed by date. The index
lists the article number, article size, and article subject. The article
itself is packed into an archive file, using Phil Katz's PKZIP. The index
information is maintained in several files. There is the master index,
which contains information on all the articles posted. There is a daily
archive index. There is a monthly archive index. These three indices are
available from the server. There is also a server index that is not
available, which includes the Message-ID from the original posting, and
tracks the References line in the header as well.
This allows the archive client (That's you!) a great deal of control in
selecting articles from the archive. You can get an entire month's worth
of articles, say for March of 1991 by sending the command:
send Mar91
You can request articles from specific dates:
send 13Mar91
send 16Mar91
send 12Jun91
You can request single articles:
send 1025
send <429@erb1.engr.wisc.edu>
Note that the article number is assigned by MY machine, not by yours.
You can also get just the indices for the first two examples:
send Mar91.idx
send 13Mar91.idx
All archive requests should be sent to REQUEST@THEREV.UUCP. Yes, I know
I've said that before. I HATE having to search through an entire
mail-server document to find the server's address.
COMMENTS AND QUESTIONS
Comments, questions, and suggestions for improving the archives or mail
server should be sent to archive@therev.uucp. Note that is NOT
request@therev.uucp.
FUTURE DIRECTIONS
The next "improvement" to the server will be (I hope) a thread
extractor, which will permit a client to request all or part of an ongoing
discussion.
---------------------------------------------------------------------
Mark A. Foster - Keeper of the Rec.Hunting archives
Archive requests to: request@therev.UUCP
Administrative notes to: archive@therev.UUCP
For not-so-smart mailers:
bbx!therev!{request | archive}@unmvax.cs.unm.edu
---------------------------------------------------------------------