[7658] in www-talk@info.cern.ch
RE: Web Scripting Languages
daemon@ATHENA.MIT.EDU (Fisher Mark)
Wed Mar 15 08:31:17 1995
Date: Wed, 15 Mar 1995 08:23:06 +0500
Errors-To: procmaster@www19.w3.org
Reply-To: FisherM@is3.indy.tce.com
From: Fisher Mark <FisherM@is3.indy.tce.com>
To: Multiple recipients of list <www-talk@www10.w3.org>
Some responses:
1) I wrote in <2F5B1459@MSMAIL.INDY.TCE.COM>:
>One possible base for this work would be Safe-Tcl, Nathaniel Borenstein's
>and Marshall Rose's email scripting extension for John Ousterhout's Tcl/Tk.
> Safe-Tcl uses a two-level interpreter, where the outer interpreter
supports
>a carefully limited set of high-level capabilities.
It is a twin interpreter, with an untrusted interpreter that runs the
Safe-Tcl scripts, while the other, trusted, interpreter can be used to
extend the untrusted interpreter. This is what I get for working from
memory on a paper read months ago...
2) Mike Meyer wrote in <19950306.798A628.667C@contessa.phone.net>:
>> i) Be able to walk the Web on their own (travel from machine to machine);
>
>Robots or spiders, which have already been written using library
>facilities as discussed above.
Robots and spiders run on a host machine but operate on a set or series (or
graph or ...) of other machines. The proposed scripts would be able to move
from machine to machine. Rather than this model (the spider/robot model):
Script on machine A
Operate on machine B
Operate on machine C
Operate on machine D
...
this model (the autonomous agent model):
Script on machine B, operating on machine B
Script on machine C, operating on machine C
Script on machine D, operating on machine D
...
could also be used for these scripts with either sequential or parallel
execution of "Script" on B, C, D, .... The first model is fine for when
"avail. CPU cycles << network bandwidth" whereas the second model is
appropriate for "avail. CPU cycles >> network bandwidth", while it's a
toss-up when available CPU cycles are close to the network bandwidth. As
Nathaniel Borenstein wrote:
>I've argued for a procedural language (safe-tcl or something
>better, if it comes along) because I want people to be able to do the
>maximal possible number of things safely. It isn't that I'm not sure
>what *I* want to do, it's that I am absolutely sure that nobody knows
>what *everybody* will want to do. For that reason, my focus has been on
>providing the maximum amount of expressive power that is compatible with
>a safe language for untrusted scripts.
The autonomous agent model at the least provides a second method for
performing Web operations that may better fit the available resources (CPU
cycles, network bandwidth, ...). There are cases (network bandwidth
constraints again, for one) where a locally executing agent might be allowed
greater permissions than a remotely executing spider.
3) Mike Meyer also wrote:
>> I suggest the name "Spider" for this Safe-Tcl extension.
>
>That name is already in use as an alias for web-wondering robots.
Agreed. How about "TclWeb", then?
======================================================================
Mark Fisher Thomson Consumer Electronics
fisherm@indy.tce.com Indianapolis, IN
"Just as you should not underestimate the bandwidth of a station wagon
traveling 65 mph filled with 8mm tapes, you should not overestimate
the bandwidth of FTP by mail."
