Subject: Re: EROS and liability
From: Ian Lance Taylor <ian@airs.com>
Date: 30 Jun 1999 11:57:36 -0400

   From: shapj@us.ibm.com
   Date: Wed, 30 Jun 1999 11:28:41 -0400

   Relatively few people hack the code directly.  A comparatively larger number
   will hack the code by applying patches or additions distributed by others.  A
   "patch", in this sense, might be the output of a diff program or it might be a
   rpm file that does a version update on a libarary **that the user does not know
   is integrity critical.**

My mind is boggling at the thought of some organization which produces
life critical devices and has such sloppy source control procedures.
When producing a life critical device, you should very carefully
examine anything you get from the original manufacturer.  You
shouldn't even consider installing unapproved third party changes.
You must be able to verify that the sources you examined are the ones
going into the device.

I think you should contact some organizations which produce life
critical devices and see what their software procedures are.

I agree that widely distributed free software makes this problem
worse, but I don't agree that this problem is unique to free software.
Similar issues arise with proprietary software if I hire a consultant
who has worked with the software at other companies.  How do I know
that he or she didn't install some patch?

   If I could come up with license wording that restricted the controls to these
   cases, I would.

What if you simply make the license wording mention life critical
devices?  Lots of existing licenses do.  I can't tell if that would
satisfy your needs or not.

Ian