Licensing of tests

[Gervase Markham]

A case has been made for changing our licensing policy such that _tests_
authored by the Mozilla community are licensed by default under CC0,
rather than the current default which is MPL 2:

https://bugzilla.mozilla.org/show_bug.cgi?id=788511

Ted Mielczarek writes:
"Test files are unique in that they're not part of what we ship, but
also can be valuable to outside parties (other browser developers,
standards organizations). They're usually standalone and simple enough
that nobody ought to care what the license is, so using public domain
feels like a good fit. Having them default to public domain means that
if someone wants to upstream them to a standards org, or share them with
another browser vendor they don't need to worry about the licensing."

Practically speaking, we would set an implementation date after which
this policy would come into effect. Tests with an explicit licensing
declaration would continue to be bound by that. We would encourage but
not require people to use the CCO declaration. But many tests are and
will continue to not have an explicit license. If someone wanted to know
the license of an unlicensed test, they could look at its checkin date.

If you object to this policy change, let me know.

Follow-ups to mozilla.legal, please.

Gerv

[Till Schneidereit]

FWIW, this is already the common practice for SpiderMonkey tests. Having it
be the explicit default would be very nice there, too, though.

[Ehsan Akhgari]

This sounds awesome!

Out of curiosity, what would this mean for tests that are already
declared to be in the public domain?
<http://mxr.mozilla.org/mozilla-central/search?string=public%20domain>

> _______________________________________________
> dev-planning mailing list
> dev-pl...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-planning
>

[Mike Hommey]

On Wed, Sep 03, 2014 at 06:21:37PM -0400, Ehsan Akhgari wrote:
> This sounds awesome!
>
> Out of curiosity, what would this mean for tests that are already declared
> to be in the public domain?
> <http://mxr.mozilla.org/mozilla-central/search?string=public%20domain>

Considering the legal uncertainty[1] of "Any copyright is dedicated to the
Public Domain", I'd say we should move them to CC0.

Mike

1. http://en.wikipedia.org/wiki/Wikipedia:Granting_work_into_the_public_domain

[Ted Mielczarek]

On 9/3/2014 7:44 PM, Mike Hommey wrote:
> On Wed, Sep 03, 2014 at 06:21:37PM -0400, Ehsan Akhgari wrote:
>> This sounds awesome!
>>
>> Out of curiosity, what would this mean for tests that are already declared
>> to be in the public domain?
>> <http://mxr.mozilla.org/mozilla-central/search?string=public%20domain>
> Considering the legal uncertainty[1] of "Any copyright is dedicated to the
> Public Domain", I'd say we should move them to CC0.
>

AFAICT those are all trying to use CC0, so if the license header is
inadequate then we should fix that.

-Ted

[Nicholas Nethercote]

On Wed, Sep 3, 2014 at 6:11 AM, Gervase Markham <ge...@mozilla.org> wrote:
> A case has been made for changing our licensing policy such that _tests_
> authored by the Mozilla community are licensed by default under CC0,

In case it's not clear (I had to look it up) CC0 is the Creative
Commons public domain dedication:
http://creativecommons.org/publicdomain/zero/1.0/

Sounds like a fine idea.

Nick

[Andreas Tolfsen]

Gervase Markham <ge...@mozilla.org>:

> A case has been made for changing our licensing policy such that _tests_
> authored by the Mozilla community are licensed by default under CC0,
> rather than the current default which is MPL 2:

It would be hugely beneficial to the project if we potentially could
take any test and submit it to the W3C's web-platform-tests repository.

If a test is in “the public domain” (CC0) my understanding is that
it's fine for them to be picked up and relicensed under the 3-clause
BSD/W3C license that w-p-t demands?

http://www.w3.org/Consortium/Legal/2008/04-testsuite-copyright.html

[Ms2ger]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Gerv,

On 09/03/2014 03:11 PM, Gervase Markham wrote:
> A case has been made for changing our licensing policy such that
> _tests_ authored by the Mozilla community are licensed by default
> under CC0, rather than the current default which is MPL 2:
>

Thanks for following up on this; I strongly support this proposal.

Ms2ger

-----BEGIN PGP SIGNATURE-----

iQEcBAEBAgAGBQJUCA0uAAoJEOXgvIL+s8n2fSkH/AztXMRpKmENB50C2TUORH5+
Nd/JponJZ3n3AiODT33cPiQzIkthSEpVCXzhTi7FdNXhK+G1dPKshsyYGZdV+cZl
ssiiW3bQ3FeWS+uV+vbYzV9t1NxtWL/wpMI5j2i39PNn2KASZbjiuE1q8JMOTX/Z
Af4eEcYjb+eSNH6sUu0CdXECvwQRAdfVUN1SunS7+PcNIpi/aZqBNVulP9FSg5SV
xgfaa0o9HlSFUDyHFiIF+MxPmmbQ96zTqBVEEL9kwkAHOWHpCzGwjuBXIhbIZJvn
QSQIYx85r9Ei/9tTTA1Xe+rUpAKZPo4SCWO0UZUwxO08Mqn13OJ6Wda+M/72ghs=
=XaPr
-----END PGP SIGNATURE-----

[Anne van Kesteren]

On Thu, Sep 4, 2014 at 4:51 AM, Andreas Tolfsen <a...@mozilla.com> wrote:
> It would be hugely beneficial to the project if we potentially could
> take any test and submit it to the W3C's web-platform-tests repository.

Yes that works. It would be even better if W3C did not change the
license if it was more liberal than theirs so changes can be imported
back without affecting the license.


--
http://annevankesteren.nl/

[Gervase Markham]

On 03/09/14 23:21, Ehsan Akhgari wrote:
> Out of curiosity, what would this mean for tests that are already
> declared to be in the public domain?
> <http://mxr.mozilla.org/mozilla-central/search?string=public%20domain>

Nothing. They would continue to be so.

Gerv

[Gervase Markham]

On 04/09/14 00:44, Mike Hommey wrote:
> Considering the legal uncertainty[1] of "Any copyright is dedicated to the
> Public Domain", I'd say we should move them to CC0.
>

> 1. http://en.wikipedia.org/wiki/Wikipedia:Granting_work_into_the_public_domain

If you look at the second line of each of those files, you'll see that
this is precisely what they are. That's the standard Mozilla CC0 header:
http://www.mozilla.org/MPL/headers/

/* Any copyright is dedicated to the Public Domain.
* http://creativecommons.org/publicdomain/zero/1.0/ */

Gerv

[Ehsan Akhgari]

Ah, indeed, you are right. I was confused because the first result from
grep (aclocal.m4) doesn't have that second line, but based on a cursory
look, that file and its copy in js/src seem to be the only ones!

Cheers,
Ehsan

[Henri Sivonen]

On Wed, Sep 3, 2014 at 4:11 PM, Gervase Markham <ge...@mozilla.org> wrote:
> Ted Mielczarek writes:
> "Test files are unique in that they're not part of what we ship, but
> also can be valuable to outside parties (other browser developers,
> standards organizations). They're usually standalone and simple enough
> that nobody ought to care what the license is, so using public domain
> feels like a good fit. Having them default to public domain means that
> if someone wants to upstream them to a standards org, or share them with
> another browser vendor they don't need to worry about the licensing."

I support this reasoning and support a policy that puts tests we write
under CC0. I think we should still permit importing tests that are
more restrictively licensed from elsewhere and allow people who write
tests as work-for-hire for the Mozilla Group to contribute to more
restrictively licensed upstream test repos under the licenses of those
repos. (Maybe the previous sentence is obvious, but I think it's worth
saying explicitly.)

> Practically speaking, we would set an implementation date after which
> this policy would come into effect. Tests with an explicit licensing
> declaration would continue to be bound by that. We would encourage but
> not require people to use the CCO declaration. But many tests are and
> will continue to not have an explicit license. If someone wanted to know
> the license of an unlicensed test, they could look at its checkin date.

I think it would make sense for tests that have been written as
work-for-hire to the Mozilla Group or have otherwise had their
copyright transferred to the Mozilla Foundation prior to the policy
change date to be relicensed by the Mozilla Foundation under CC0. That
way, if a test predated the policy change but the logs showed it was
authored by an Mozilla employee or contractor, you could treat it as
CC0 without further bureaucracy.

--
Henri Sivonen
hsiv...@hsivonen.fi
https://hsivonen.fi/

[Gervase Markham]

On 05/09/14 10:01, Henri Sivonen wrote:
> I support this reasoning and support a policy that puts tests we write
> under CC0. I think we should still permit importing tests that are
> more restrictively licensed from elsewhere

As long as they meet our license policies, sure. (For code which is not
shipped, the official policy is simply "must be under an OSI license",
and the unofficial policy is also "should not cause us licensing or
administrative hassle".)

> and allow people who write
> tests as work-for-hire for the Mozilla Group to contribute to more
> restrictively licensed upstream test repos under the licenses of those
> repos.

Again, if open source, no problem; if not, we'd have to consider on a
case-by-case basis. We'd be more likely to establish an independent test
suite.

> I think it would make sense for tests that have been written as
> work-for-hire to the Mozilla Group or have otherwise had their
> copyright transferred to the Mozilla Foundation prior to the policy
> change date to be relicensed by the Mozilla Foundation under CC0. That
> way, if a test predated the policy change but the logs showed it was
> authored by an Mozilla employee or contractor, you could treat it as
> CC0 without further bureaucracy.

I'm happy to do that if all it means is stating it as a policy, but I
don't have time to go and do the research and change license headers.

We can simply say: "if you do the research on a file with no header and
find it qualifies, you can use it under CC0. If you can't be bothered,
you must assume it's MPL2."

Gerv

[Henri Sivonen]

On Fri, Sep 5, 2014 at 12:16 PM, Gervase Markham <ge...@mozilla.org> wrote:
> On 05/09/14 10:01, Henri Sivonen wrote:

>> I support this reasoning and support a policy that puts tests we write
>> under CC0. I think we should still permit importing tests that are
>> more restrictively licensed from elsewhere
>

> As long as they meet our license policies, sure. (For code which is not
> shipped, the official policy is simply "must be under an OSI license",
> and the unofficial policy is also "should not cause us licensing or
> administrative hassle".)

Sounds good.

>> and allow people who write
>> tests as work-for-hire for the Mozilla Group to contribute to more
>> restrictively licensed upstream test repos under the licenses of those
>> repos.
>

> Again, if open source, no problem; if not, we'd have to consider on a
> case-by-case basis. We'd be more likely to establish an independent test
> suite.

The suites I immediately had in mind are the htm5lib test suite (MIT
so OSI approved) and various W3C test suites, which are dual licensed
under OSI-approved 3-clause BSD and non-OSI-approved W3C Test Suite
License.

Since CC0 and permissive licenses, including BSD, permit proprietary
forks, I don't see explicitly allowing the W3C to license the tests
also under the W3C Test Suite License is worse than allowing them to
do so less explicitly via CC0 or BSD.

>> I think it would make sense for tests that have been written as
>> work-for-hire to the Mozilla Group or have otherwise had their
>> copyright transferred to the Mozilla Foundation prior to the policy
>> change date to be relicensed by the Mozilla Foundation under CC0. That
>> way, if a test predated the policy change but the logs showed it was
>> authored by an Mozilla employee or contractor, you could treat it as
>> CC0 without further bureaucracy.
>

> I'm happy to do that if all it means is stating it as a policy, but I
> don't have time to go and do the research and change license headers.

I didn't mean to suggest anyone would do the research proactively.

> We can simply say: "if you do the research on a file with no header and
> find it qualifies, you can use it under CC0. If you can't be bothered,
> you must assume it's MPL2."

This is what I meant.