- From: Sarr Blumson <sarr@citi.umich.edu>
- Date: Tue, 04 Apr 1995 09:55:15 -0400
- To: David.Halls@cl.cam.ac.uk
- Cc: sarr@citi.umich.edu, www-talk@www10.w3.org
In message <9504030915.AA12306@ouse.cl.cam.ac.uk>you write: >> >> Others have mentioned the safety issue, but let me put it a little more >> strongly. I will NEVER EVER run a browser that depends on (no, is >> willing to) executing binaries downloaded from a server, at least on >> any currently extant hardware architecture. Nor will I trust a machine >> where somebody has. > >So you will NEVER download packages from the net, compile them and install >them? You rely on your native OS and its utilities completely. No-one >checks source code (e.g. Gnu <fill-in-here>, XV etc etc) for "rm -r *". >Just because you compile them doesn't make them safe. The same amount >of trust applies. Of course I do those things. Sometimes. When I do I think carefully about where I'm getting them from, look at the source, and run them for a while under an account I keep for that purpose with no access to anything (the reason why I believe that even single user machines need multiuser security, but that's another argument). People actually do this. I recall a discussion on this very list a few months ago about a package who's installation script downloaded another script and executed it without warning. People noticed. If I were a using a browser that downloaded binary applets on a regular basis, even that level of care would become unmanageable. Even assuming that the browser warned me that it was happening. -------- Sarr Blumson sarr@umich.edu voice: +1 313 764 0253 FAX: +1 313 763 4434 CITI, University of Michigan http://www.citi.umich.edu:80/users/sarr / 519 W William, Ann Arbor, MI 48103-4943
Received on Tuesday, 4 April 1995 11:00:18 UTC