Re: How about a Safe Virtual Machine? or extensable network directories and agent functionallity

Thomas Walsh (
Wed, 5 Oct 1994 08:47:33 +0100


From: Karl Auerbach <>

> We can get part way there just by having common standards for posting
> rates, terms and conditions. (By analogy with a restaurant, that's
> like posting the menu to a restaurant outside the door. Of course it
> doesn't tell you whether the food is good, but at least you know the
> price and whether they take checks.)

I find this sort of information *from servers* to be extremely

In addition, one thing I've been cogitating on is how to express
copyright restrictions/licensing terms/payment agent information in a
way that a user's automated browser can read *before* fetching the
entire document (i.e. using a HEAD request). Making copyright
comprehensible to a *program* (especially in my self-imposed limit of
400 bytes, max) is a bit of a challenge. (But I have some ideas which
I will post later this month after they I get a chance to bounce them
off some lawyers better educated in this area than I am.)


I also have to address this issue. We should collaboratively make a list of
the elements in question, then we can better divide what needs to be done
with each element and how we as a group can handle it.

I have stated before that I think it is pertinant to be able to
automatically search a "distributed directory of information"
that would contain some of these elements. An agent, client ...
whatever could then find all relavent materials ( bounded by
some constaints ( part of the agent or client request, a multimedia
"time to live" on steroids type of functionallity :) ) and
return these to the actual user. The user software then can
intelligently (we hope!) choose what information, and at what
cost, copyright... they are willing to commit to downloading.

There is no reason HTTP has to be used for this
"directory search mechanism" because there are no "shipping"
clients or servers that can do this now so everyone will need
to do work in one form or another to make it work.
If HTTP is used we need to define the extensions.

On the Safe-Tcl side of things: Each server should be able to
advertise ( in some fashion ) to clients, agents and other servers
what "native" core functionallity "each class" of user or agent
can use. It should also advertise what set of "mobile" functionallity
it is willing to bind with (IE being sent a safe-tcl script to
do group mailings, search for keywords etc...).

We need to list core agentd functionallity required for a server.
We need to list possible extendable functionallity an agent can give.

We need to define the "class" of users and how to safely determine
they are who they are (PGP...ect, don't invent but use :)

PS: I am using safe-tcl in general here, there may be something better
but you all can relate with it whether its what ultimately gets used.

"The Information Superhighway At Your FingerTips"
#include <std.disclaim.h>