[ohf-licenses] Open Hardware baby step, APSL
Greg London
email at greglondon.com
Tue Mar 25 20:52:10 EDT 2008
OK, so the hardware license I proposed isn't getting overwhelming
responses as planned. I was sick for the last week and during the
occaisional lucide moment between consciousness and SpongeBob, I had the
idea that maybe it's too much at once, and what was needed was a baby step
that at least starts in the right direction. So, with that, I think one
step that open hardware folks could take that would be a huge step in
protecting their works for the community would be to switch from GNU-GPL
to the APPLE PUBLIC SOURCE LICENSE.
http://www.opensource.org/licenses/apsl-2.0.php
The main issue with GNU-GPL is that the copyleft aspect of the license is
triggered by distribution of the work. In hardware, once the work is
translated into silicon or printed circuit, it is no longer a copyright
derivative, and selling the parts does not count as copyright
distribution. Therefore, most hardware projects using GNU-GPL will not
receive any copyleft benefits. At All. GNU-GPL, when applied to hardware
projects, is much more like applying a MIT or BSD style license to the
work, since the rights are released, but copyleft is almost never
triggered.
The Apple Public Source License is a copyleft license that was designed to
close the loophole in the GNU-GPL that was being exploited by websites.
Because GNU-GPL requires distribution, a website could modify GNU-GPL code
and host it on a computer, and present users to the input/output streams.
Since the host software isn't distributed, the copyleft aspect of GNU-GPL
would not trigger, and the derivative could be maintained as a proprietary
fork.
The APSL closed this loophole by saying that the copyleft aspect of the
license is triggered by the creation of a copyright derivative, even if
that derivative is never distributed. Therefore, if the hosting website
creates a derivative work, it must make the derivative publicly available
under the same license as the original. Even if the derivative would
normally be hidden behind a host interface.
This is the exact same loophole that occurs in hardware. People create
hardware designs and put them under a license like GNU-GPL. Someone comes
along, creates a derivative, converts the whole thing into a piece of
silicon, and the silicon is sold, without distributing the derivative. So
the copyleft protection of GNU-GPL does not get triggered.
By using the APSL for hardware, a person can put their core under the
license, and when someone makes a derivative, that derivative must be made
publically available under the APSL.
The thing would be that the way copyright law might interpret things, any
use of a core might be considered a derivative. Therefore, at the very
minimum, it would be useful to draw a boundary at the top level and say
that anyone can use the design as a whole in a bigger project without it
being considered a derivative. But any modification within that boundary
would be treated as a derivative and would trigger the copyleft aspect of
the APSL. Otherwise, instantiating a design inside a larger design might
be considered a copyright 'derivative', and the number of users for such a
core would drop considerably.
Would this be a step in a direction that everyone could agree with?
I'll have to re-read the APSL again and see how close it is to what we
would need for a hardware license. But I thought I'd run the idea here
first just to see if it would be generally agreeable or adamantly
opposable.
Greg
More information about the ohf-licenses
mailing list