Subject: RE: For approval: Open Test License v1.1
From: Alex Rousskov <rousskov@measurement-factory.com>
Date: Thu, 8 Jan 2004 23:55:13 -0700 (MST)


On Thu, 8 Jan 2004, Lawrence E. Rosen wrote:

> > : 3. Publication of results from standardized tests contained within
> > :    this software (<TESTNAME>, <TESTNAME>) must either strictly
> > :    adhere to the execution rules for such tests or be accompanied
> > :    by explicit prior written permission of <OWNER>.
>
> You can't place requirements on the ways that downstream licensees
> can use their copies or on the types of derivative works they
> create.

Lawrence,

	The above is not meant to restrict the ways to use copies or
the types of derivative works.  It is meant to restrict only how test
results are _named_. If a user renames a standard test or invents her
own new test, she can publish whatever she wants, regardless of
standardized test rules.

Please note that OSI certified licenses already have similar (but not
generic enough)  clauses! See, for example, Artistic License and Open
Group Test Suite license. Both require users to rename standardized
tests if standardized tests are modified.

> We discussed similar things when the OSI board approved Sun's SISSL,
> and the SISSL solves this problem of test standardization.  I
> suggest you take a closer look at that license

Thanks for the pointer! I may have missed something important (the
license is rather long), but I believe the SISSL license is more
restrictive than the above clause. Specifically, Sun says:

``The Modifications which You create must comply with all requirements
set out by the Standards body...''

Open Test License does not require that. Quoting further:

``In the event that the Modifications do not meet such requirements,
You agree to publish either (i) any deviation from the Standards
protocol resulting from implementation of Your Modifications and a
reference implementation of Your Modifications or (ii) Your
Modifications in Source Code form, and to make any such deviation and
reference implementation or Modifications available to all third
parties under the same terms as this license on a royalty free basis
within thirty (30) days of Your first customer shipment of Your
Modifications.''

Open Test License does not require any of the above either!

The only thing we want to require is that if standard test is executed
in violation of standard test rules, the result does not use a
standard name to avoid misleading publications. Does that really
contradict any OSD clauses? I do not think so, and the fact that OSI
certified at least two "rename if you modify" licenses seem to support
my belief. Does it not?

I suspect that Sun holds trademarks for test names so they can enforce
naming restrictions that way (otherwise, I do not see how their
license would prevent a 3rd party from completely trashing Sun
products using a modified version of Sun's own benchmark). As I noted
in my original message, many small owners do not have the resources to
enforce trademarks and is likely to lose them. That is why we want
clause #3 there.

We could have made a more restrictive clause saying that <TESTNAME>s
shall not be used to name any published test results. That would be
equivalent to Apache saying that Apache name should not be used.
However, that would prevent people from publishing recognizable
standard results, and we want to allow that.


Do you see what we are trying to get at? Do the above clarifications
address your concerns? Do you think our intent/goal is
OSI-certifiable? We would be happy to improve the text of clause #3 to
make it match the intent better, of course.

Thank you,

Alex.

 : 3. Publication of results from standardized tests contained within
 :    this software (<TESTNAME>, <TESTNAME>) must either strictly
 :    adhere to the execution rules for such tests or be accompanied
 :    by explicit prior written permission of <OWNER>.
--
license-discuss archive is at http://crynwr.com/cgi-bin/ezmlm-cgi?3