Licensing

[Anne van Kesteren-2]

Hi,

I think it would be good if the tests were licensed under the MIT license  
as that makes it much more clear for software projects whether they can  
reuse them. I believe the W3C Document License is currently not suitable  
for everyone.

Kind regards,


--
Anne van Kesteren
<http://annevankesteren.nl/>
<http://www.opera.com/>

[fantasai]

Anne van Kesteren wrote:
>
> Hi,
>
> I think it would be good if the tests were licensed under the MIT
> license as that makes it much more clear for software projects whether
> they can reuse them. I believe the W3C Document License is currently not
> suitable for everyone.

Can you give a specific example, please?

~fantasai

[Ian Hickson]

On Tue, 8 Jan 2008, fantasai wrote:

> Anne van Kesteren wrote:
> >
> > Hi,
> >
> > I think it would be good if the tests were licensed under the MIT license as
> > that makes it much more clear for software projects whether they can reuse
> > them. I believe the W3C Document License is currently not suitable for
> > everyone.
>
> Can you give a specific example, please?

The W3C document license doesn't allow for modifications, and requires
more boilerplate than ideal. This makes it unsuitable for a number of
projects. It also is not one of the most common licenses, and would
therefore block anyone from, e.g., using these tests in a Google Code
repository. I second Anne's request that a more suitable license be used
(such as the three-clause BSD, Apache v2, or MIT licenses).

--
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

[fantasai]

Ian Hickson wrote:

> On Tue, 8 Jan 2008, fantasai wrote:
>> Anne van Kesteren wrote:
>>> Hi,
>>>
>>> I think it would be good if the tests were licensed under the MIT license as
>>> that makes it much more clear for software projects whether they can reuse
>>> them. I believe the W3C Document License is currently not suitable for
>>> everyone.
>> Can you give a specific example, please?
>
> The W3C document license doesn't allow for modifications, and requires
> more boilerplate than ideal. This makes it unsuitable for a number of
> projects.

Yes, this I agree is a problem, and I'm talking with Rigo about it.
But again, if you or Anne have specific examples that would help.

~fantasai

[Ian Hickson]

On Wed, 9 Jan 2008, fantasai wrote:
>
> Yes, this I agree is a problem, and I'm talking with Rigo about it. But
> again, if you or Anne have specific examples that would help.

It would presumably be a problem for Microsoft, Opera, Mozilla, and Apple,
for a start, since all four occasionally have to modify tests as part of
incorporating them into their test harnesses.

--
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

On Wednesday 09 January 2008 13:19, Ian Hickson wrote:
> On Wed, 9 Jan 2008, fantasai wrote:
> > Yes, this I agree is a problem, and I'm talking with Rigo about it.
> > But again, if you or Anne have specific examples that would help.
>
> It would presumably be a problem for Microsoft, Opera, Mozilla, and
> Apple, for a start, since all four occasionally have to modify tests
> as part of incorporating them into their test harnesses.

Test suite licenses have been discussed a lot in W3C over the years. The
current policy[1] is the result. The fact that it has been stable since
at least 2004 indicates to me that we need more than presumptions
before discussing it again.

It seems to me that conformance isn't helped by allowing people to make
modifications to the test suite without review by the WG.

The current license requires that somebody who wants to publish modified
tests asks permission. (Making and using them is fine.) If W3C gets
overwhelmed with requests it is early enough to think about a new
system. The few requests we got so far didn't pose problems.

[1] http://www.w3.org/2004/10/27-testcases



Bert
--
  Bert Bos                                ( W 3 C ) http://www.w3.org/
  http://www.w3.org/people/bos                               W3C/ERCIM
  bert@...                             2004 Rt des Lucioles / BP 93
  [phone]            06902 Sophia Antipolis Cedex, France

[Ian Hickson]

On Wed, 9 Jan 2008, Bert Bos wrote:
>
> Test suite licenses have been discussed a lot in W3C over the years. The
> current policy[1] is the result. The fact that it has been stable since
> at least 2004 indicates to me that we need more than presumptions before
> discussing it again.

It's been stable mostly because there aren't many test suites, and few
people use those that exist. And it's not clear that people are actually
obeying the license in the cases where they do. (The DOM test suite uses a
different license, the W3C software license, which is slightly less
restrictive and still, as far as I can tell, gets violated almost every
time that test suite is used.)


> It seems to me that conformance isn't helped by allowing people to make
> modifications to the test suite without review by the WG.

Modifications are frequently required to adapt test cases to automated
test harnesses.


> The current license requires that somebody who wants to publish modified
> tests asks permission. (Making and using them is fine.)

It's unclear how this works in the open source world. If contributor A
takes the tests and gets permission to change them and publish the
changes, and commits those tests, and contributor B then publishes the
tarball, has B got permission to do this? What if the product forks? Is
this even compatible with the more common open source licenses? What if
someone wants to check the test suite into a system that only accepts the
mainstream open source licenses, like Google Code?


To make the issue simple: I refuse to license my tests under the W3C
Document License or the W3C Software License. I would be happy to licenes
my tests under the 3-clause BSD, MIT, or Apache v2 licenses. There is
precedent in the W3C for this; the HTML working group's tests are to be
licensed under the MIT license.

The W3C shouldn't be in the business of encouraging the proliferation of
software licenses. It's a big enough mess as it is.

--
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

[Alex Mogilevsky]

I am not a lawyer but I think you get it backwards. It is your work, then you get to choose how to license it. If you license it to W3C which then publishes your work under a more restrictive license (I am assuming you are not happy about that part), the original work is still available from you directly, isn't it?

-----Original Message-----
From: public-css-testsuite-request@... [mailto:public-css-testsuite-request@...] On Behalf Of Ian Hickson
Sent: Wednesday, January 09, 2008 6:35 PM

To make the issue simple: I refuse to license my tests under the W3C
Document License or the W3C Software License. I would be happy to licenes
my tests under the 3-clause BSD, MIT, or Apache v2 licenses. There is
precedent in the W3C for this; the HTML working group's tests are to be
licensed under the MIT license.

[Ian Hickson]

On Wed, 9 Jan 2008, Alex Mogilevsky wrote:
>
> I am not a lawyer but I think you get it backwards. It is your work,
> then you get to choose how to license it. If you license it to W3C which
> then publishes your work under a more restrictive license (I am assuming
> you are not happy about that part), the original work is still available
> from you directly, isn't it?

Yes. I'm saying that I would not be willing to license my tests to the W3C
under a different license than the 3-clause BSD, MIT, or Apache v2 licenses.

--
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

[L. David Baron]

On Wednesday 2008-01-09 18:56 -0800, Alex Mogilevsky wrote:
> I am not a lawyer but I think you get it backwards. It is your work,
> then you get to choose how to license it. If you license it to W3C
> which then publishes your work under a more restrictive license (I am
> assuming you are not happy about that part), the original work is
> still available from you directly, isn't it?

I think you're confusing licensing and copyright assignment.  A
copyright holder, or all of them if there are multiple, can offer their
works under additional licenses.  Some open-source projects require
assignment of copyright so that a single entity has this power; some
projects leave the copyright with the author and accept work offered
under a given license or set of licenses.

(I am not a lawyer, etc.)

-David

--
L. David Baron                                 http://dbaron.org/
Mozilla Corporation                       http://www.mozilla.com/

Ian,

> On Wed, 9 Jan 2008, Alex Mogilevsky wrote:
> > I am not a lawyer but I think you get it backwards. It is your
> > work, then you get to choose how to license it. If you license it
> > to W3C which then publishes your work under a more restrictive
> > license (I am assuming you are not happy about that part), the
> > original work is still available from you directly, isn't it?
>
> Yes. I'm saying that I would not be willing to license my tests to
> the W3C under a different license than the 3-clause BSD, MIT, or
> Apache v2 licenses.

By doing so, you're not compliant to any of the licenses you mention
anymore. So Google would have to refuse your tests too.

Best,

Rigo

[Lachlan Hunt]

Rigo Wenning wrote:
> Ian Hickson wrote:
>> Yes. I'm saying that I would not be willing to license my tests to
>> the W3C under a different license than the 3-clause BSD, MIT, or
>> Apache v2 licenses.
>
> By doing so, you're not compliant to any of the licenses you mention
> anymore. So Google would have to refuse your tests too.

While those 3 licences would allow the tests to be used in projects with
alternative licences [1], they do not allow them to be re-licensed under
a different licence, which, if I understand correctly, the W3C would do
if they included the tests in the offical test suite.

[1] Note that that Apache v2 is incompatible with GPLv2, so I wouldn't
     recommend using that one.

--
Lachlan Hunt - Opera Software
http://lachy.id.au/
http://www.opera.com/

[fantasai]

Ian Hickson wrote:
> On Wed, 9 Jan 2008, Alex Mogilevsky wrote:
>> I am not a lawyer but I think you get it backwards. It is your work,
>> then you get to choose how to license it. If you license it to W3C which
>> then publishes your work under a more restrictive license (I am assuming
>> you are not happy about that part), the original work is still available
>> from you directly, isn't it?
>
> Yes. I'm saying that I would not be willing to license my tests to the W3C
> under a different license than the 3-clause BSD, MIT, or Apache v2 licenses.

This is not being very helpful. I can add your tests right now under both the
BSD, MIT, or Apache license (or all three) and the W3C Document License grant,
but I cannot add them if you refuse to license them under the W3C license. If
these are the terms of your contribution, you should have mentioned that in
your message in September.
   http://lists.w3.org/Archives/Public/public-css-testsuite/2007Sep/0010.html

Arron Eischolz has already put in many hours reviewing your tests to prep
them for adding to the test suite. I don't have a problem with you imposing
such constraints on any new tests, but I think it would be rather vicious of
you to retract your existing offer like this. It's not like you didn't know
the licensing policy when you made it.

~fantasai

[L. David Baron]

On Wednesday 2008-01-09 17:13 +0100, Bert Bos wrote:
> The current license requires that somebody who wants to publish modified
> tests asks permission. (Making and using them is fine.) If W3C gets
> overwhelmed with requests it is early enough to think about a new
> system. The few requests we got so far didn't pose problems.

That's not sufficient for most open-source projects.  Open source
projects typically provide their entire source under a specific
license or set of licenses.  They don't provide it under the terms
"you can use it under license X if you ask the permission of the
following 15 companies/people/organizations".  They provide it under
"you can use it under license X".  The difference between those two
can be a major obstacle to such use.  (Given the chance that at
least 1 of the 15 organizations won't respond, or will deny the
request, it's usually not even worth bothering to ask.)

-David

--
L. David Baron                                 http://dbaron.org/
Mozilla Corporation                       http://www.mozilla.com/

David Baron wrote:

> On Wednesday 2008-01-09 17:13 +0100, Bert Bos wrote:
> > The current license requires that somebody who wants to publish
> > modified tests asks permission. (Making and using them is fine.)
> > If W3C gets overwhelmed with requests it is early enough to think
> > about a new system. The few requests we got so far didn't pose
> > problems.
>
> That's not sufficient for most open-source projects.  Open source
> projects typically provide their entire source under a specific
> license or set of licenses.  They don't provide it under the terms
> "you can use it under license X if you ask the permission of the
> following 15 companies/people/organizations".  They provide it
> under "you can use it under license X".  The difference between
> those two can be a major obstacle to such use.  (Given the chance
> that at least 1 of the 15 organizations won't respond, or will deny
> the request, it's usually not even worth bothering to ask.)

Neither a testsuite nor a specification trying to achieve
interoperability is an open-source project. It has different social
semantics. Do you think you can have a FIPS-140 certification that
you can build yourself and everybody can just alter and run?
Testsuites are the first step of certification. What if everybody
could invent their own ISO 9002 certification? Code is different. The
fact that some code is used in testsuites does not mean it has the
same goal and finality as a browser or a word processor. Copyright is
used and meant to protect this kind of activity. One can do some
stuff with trademark law, but this would mean a complete reset of the
current W3C system with consequences for us all that I cannot assess
here and within the given timeframe.

BTW: "you can use under license X" is also true for the Document
license as the document license allows for "any use", so it does not
describe well what you mean. I think I understood what you mean and
this is the social process. BTW, you even MUST ask the following 15
companies/people/organizations to get your patch back into the
source. The social control just happens to be on a different level.

So Bert is right, lets wait for the requests and think about new
solutions if the old ones don't work anymore.

Best,

Rigo

[Ian Hickson]

On Thu, 10 Jan 2008, Rigo Wenning wrote:
>
> Neither a testsuite nor a specification trying to achieve
> interoperability is an open-source project.

No, but test suites are imported into open source projects and embedded
into automated test systems as part of open source projects, and thus need
to fulfill the conditions needed for open source projects. (This also
applies to closed-source projects, though it is not an issue in this
instance since they don't have to republish the tests.)


> Do you think you can have a FIPS-140 certification that you can build
> yourself and everybody can just alter and run?

These are not certification test suites.


> So Bert is right, lets wait for the requests and think about new
> solutions if the old ones don't work anymore.

We have three requests now, one from Anne (Opera), one from me (Google),
and one from David (Mozilla). How many more would you like?


On Thu, 10 Jan 2008, Rigo Wenning wrote:
> >
> > Yes. I'm saying that I would not be willing to license my tests to the
> > W3C under a different license than the 3-clause BSD, MIT, or Apache v2
> > licenses.
>
> By doing so, you're not compliant to any of the licenses you mention
> anymore. So Google would have to refuse your tests too.

This is incorrect, and shows a somewhat fundamental misunderstanding of
copyright and license law.

--
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

[L. David Baron]

On Thursday 2008-01-10 22:01 +0100, Rigo Wenning wrote:
> Neither a testsuite nor a specification trying to achieve
> interoperability is an open-source project. It has different social

But it would be good if the test suites were used by open source
projects for things like regression testing.  This often requires:

 1) slight or substantial modifications to the tests to fit into a
 regression test framework

 2) permission automatically being granted to any contributor to the
 project under a license equally or less restrictive than that
 project's license, so that anybody can participate in the open
 source project as a tester/developer without asking permission
 (e.g., to improve the test harness, fix bugs in tests, add new
 tests that are variants of old ones, etc.).

> BTW: "you can use under license X" is also true for the Document
> license as the document license allows for "any use", so it does not

No it doesn't, since one of the uses "license X" allows is
modification (if it's an open-source license).

For an open-source project to use tests, those tests would need to
be licensed under an appropriate open-source license, which the W3C
document license is not.  If the W3C is willing to license its test
suites under such a license, it should do so rather than saying "ask
us if you want us to do so".  Saying "ask us" doesn't make any sense
since what's needed is a license to *anybody*.  Or, to put it
another way, we're asking you now.


I think the position you're stating effectively means that open
source implementations of standards shouldn't participate in W3C
test suite development, but should instead pool their resources to
work on their own test suite at a different organization with
acceptable licensing.

-David

--
L. David Baron                                 http://dbaron.org/
Mozilla Corporation                       http://www.mozilla.com/

[Ian Hickson]

On Thu, 10 Jan 2008, fantasai wrote:
>
> This is not being very helpful. I can add your tests right now under
> both the BSD, MIT, or Apache license (or all three) and the W3C Document
> License grant, but I cannot add them if you refuse to license them under
> the W3C license.

I want the whole test suite licensed under BSD or MIT (or Apache v2), not
just my tests.


> If these are the terms of your contribution, you should have mentioned
> that in your message in September.
>   http://lists.w3.org/Archives/Public/public-css-testsuite/2007Sep/0010.html

At the time I was not aware that the W3C had such a ridiculous policy.
Indeed, I would have insisted we change the license on the Selectors test
suite many years ago if I had realised the W3C's policy. I only became
aware of the issue at the HTML working group F2F meeting, and since that
test suite was quickly resolved to be covered by the MIT license, I had no
worries that there would be a problem with the CSS tests. Clearly I was
mistaken.


> Arron Eischolz has already put in many hours reviewing your tests to
> prep them for adding to the test suite.

Incidentally, I would recommend anyone reviewing or working with these
tests to contact me, so that I can keep their needs in mind. Did Arron
find any problems that I should resolve?

--
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Note that the MIT license and the BSD license allow for derivatives
and their licensing under any license provided the notice is
provided. They have no viral effect as the W3C Software license.

Best,

Rigo

On Thursday 10 January 2008, Lachlan Hunt wrote:
> Rigo Wenning wrote:
> While those 3 licences would allow the tests to be used in projects
> with alternative licences [1], they do not allow them to be
> re-licensed under a different licence, which, if I understand
> correctly, the W3C would do if they included the tests in the
> offical test suite.
>
> [1] Note that that Apache v2 is incompatible with GPLv2, so I
> wouldn't recommend using that one.

On Thursday 10 January 2008, Ian Hickson wrote:

> On Thu, 10 Jan 2008, Rigo Wenning wrote:
> > Neither a testsuite nor a specification trying to achieve
> > interoperability is an open-source project.
>
> No, but test suites are imported into open source projects and
> embedded into automated test systems as part of open source
> projects, and thus need to fulfill the conditions needed for open
> source projects. (This also applies to closed-source projects,
> though it is not an issue in this instance since they don't have to
> republish the tests.)

I see the issue. BTW, the W3C Software license was recognized as being
compatible with the GPL:
http://www.w3.org/Consortium/Legal/IPR-FAQ-20000620#GNU

This issue is not easy to resolve as it invokes a conflict of
different interests. We'll have to talk to those people and see how
we can resolve that.

>
> > Do you think you can have a FIPS-140 certification that you can
> > build yourself and everybody can just alter and run?
>
> These are not certification test suites.

What are the semantics of those test suites? What do they assert
socially? In all other Groups, tests are contributed back to the
Group's test suite. And there are many open source projects involved.

>
> > So Bert is right, lets wait for the requests and think about new
> > solutions if the old ones don't work anymore.
>
> We have three requests now, one from Anne (Opera), one from me
> (Google), and one from David (Mozilla). How many more would you
> like?

I do not consider three requests overwhelming.

> On Thu, 10 Jan 2008, Rigo Wenning wrote:
> > By doing so, you're not compliant to any of the licenses you
> > mention anymore. So Google would have to refuse your tests too.
>
> This is incorrect, and shows a somewhat fundamental
> misunderstanding of copyright and license law.

...says the lawyer

Best,

Rigo