From: stewarts@ix.netcom.com
Date: Sat, 31 Aug 1996 18:40:51 +0800
To: "Osborne, Rick" <OSBORRI@msmail.northgrum.com>
Subject: Re: Real-time key server
Message-ID: <199608310844.BAA00275@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


At 04:30 PM 8/27/96 DST, "Osborne, Rick" <OSBORRI@msmail.northgrum.com> wrote:
>In my research into keyservers, I find that none of them are realtime. 
> Sure, there are CGI interfaces to them, but that's not what I mean.  I was 
>hoping for something along the line of a Finger or SMTP protocol, such that 
>a client connects to a host (say, keyserv.northgrum.com) on a certain port 
>(say, 5397) and goes through a realtime version of what is accomplished via 
>email. [...]
>Why?  Because:
>1. I do not want to reinvent the wheel.
>2. I would like to conform to existing standards.

There are at least three shapes of wheel readily available.

Some of the keyservers have had finger interfaces; there was one on
wasabi.io.com for a while, though it may be inactive.  John Perry
may know where to find source code for it?  Assuming you want PGP keys
delivered in ASCII format, finger protocol should work just fine - you could
easily enough hack your own fingerd server to live on the fingerd port,
which gets PGP key files from a directory or database instead of from
users' home directories, and run it on pgp.northgrum.com or whatever.
You could even hack the existing finger code from {Free,Net}BSD,
making sure to clean up any remaining sprintf()s and {*}get()s.

Alternatively, a CGI query _is_ an easy way to do it, and there's
existing code on the MIT keyserver that's blazingly fast.

John Gilmore's SWAN project is using a DNS-based PGP key server;
check out http://www.cygnus.com/~gnu/swan.html for more details.


>Okay, I was told to not post to the list until I'd:
>1. Lurked for a month,
>2. Figured out who Detweiler was,
>3. Found out about BlackNet and DC Nets,
>4. Learned of at least three of David Chaum's innovations.
>And at that point "[I] may be ready to post [my] first comments."
>
>Well, dammit, I can't wait that long.  I need help now.  I've only been on 
>for 2 weeks, I have no clue who Detweiler is (other than the
>welcome message I haven't heard two words about him/her/it), and I know 
>enough to sound stupid about the other stuff.  But if you still think I'm 
>worthy of listeneing to, then read on.

Three out of four ain't bad, and Detweiler's been using his alias on
the list for a couple of years and acting quite civilized :-)
Some good reading sources are Bruce Schneier's book "Applied Cryptography"
and Tim May's "Cyphernomicon" piece which is on the web somewhere.
Also, asking for pointers to existing work does categorize you
with the clueful minority.....

#			Thanks;  Bill
# Bill Stewart, +1-415-442-2215 stewarts@ix.netcom.com
# <A HREF="http://idiom.com/~wcs"> 	Reassign Authority!