From: Jason W Solinsky <solman@MIT.EDU>
Date: Sun, 21 Aug 94 20:58:47 PDT
To: Jim Hart <hart@chaos.bsu.edu>
Subject: Re: www log stripper
In-Reply-To: <199408220158.UAA16851@chaos.bsu.edu>
Message-ID: <9408220358.AA09407@ua.MIT.EDU>
MIME-Version: 1.0
Content-Type: text/plain


> First question: does anybody know the full URL for detailed
> information on how World Wide Web servers do logging, and
> the particular kinds of connection  and transaction information
> they can log, and what information they typically do log?
> (I know the main Mosaic doc URLs but can't find this info).

It depends on the server. My log files include something about every
single request made to the server. Since I am doing marketing, I can
then search through these and based on the access times, host names
and domains, document accesses, and information supplied by the users
using forms determine what changes need to be made and what follow up
is appropriate for each person.

If you want anonymity use a proxy.

> Second question: wev'e heard about installing swIPe in the 
> kernel and using it as an anonymous packet forwarder -- stripping 
> off the original.  Has anybody done the analogous thing at the 
> http protocol level for the World Wide Web?  That is, set up an
> httped, the only purpose of which is to forward URL requests with all
> originating site and username information stripped.  I would
> love to use such a service.  I would even pay substantial amounts
> of "Magic Money" tokens to do so, if somebody writes a convenient 
> user interface to that system.   Also, if some folks are serious
> about taking the lead on either of these projects, I'd be
> happy to contribute my hacking skills (the typical C/Berkeley
> Unix & networking, etc.) and do beta testing.

I'm debuging a product that will do this. Despite an effort to write
extremely clean code, I seem to have a number of problems and I can't
say when it will be ready even for alpha testing. :(

JWS