Subject: Re: EROS license
From: Ian Lance Taylor <ian@airs.com>
Date: 28 Jun 1999 15:39:14 -0400

   From: shapj@us.ibm.com
   Date: Mon, 28 Jun 1999 15:17:02 -0400

   >   As someone who wants to use a dual use license, I have a liability exposure.
   >    ...
   >   Can anyone suggest a practical solution that won't have me in court every
   time
   >   somebody in the open software community feels like it?
   >
   >I don't see anything new about this, nor any need to worry about the
   >sophistication of the courts.  Publishers and authors face these
   >issues regularly, when somebody claims ``that book is clearly based on
   >the one I submitted to you two years ago.''

   The difference is that when your survivors put me in front of a jury after your
   pacemaker fails and puts a regrettably premature end to our exchange, what the
   jury will see is that the pacemaker ran EROS, not that the Fred Flintstone patch
   was responsible for the problem.

   Thus, in practice, liability for patches can well fall on the originator.

I've already said that I don't think the GPL works for embedded
systems, and in fact I mentioned this same liability problem on the
list yesterday.

I think that in this case the party to blame is the manufacturer of
the pacemaker, who should not have permitted anybody to change their
software (and hence should not have used any software under the GPL).

If the pacemaker manufacturer used a modified version of your
software, I think you can easily explain that they did not use your
certified version, and that you are not at fault.  This is no
different from supplying hardware which only functions under certain
conditions; if the manufacturer doesn't follow the guidelines, it's
their fault, not the suppliers.  Moreover, most kernel distributors do
after all provide source code, albeit code the customer is not
permitted to distribute, so I don't see why using a free software
license introduces any additional liability.

Ian