Revisit LICENSE.md?

[plehegar]

The current LICENSE.md comes from history in W3C around test suites and hasn't been looked at in the past 4 years at least.

We should look at it again and see if it still makes sense. If not, we need to figure out what would make sense. and then a transition path from the current one to a new one.

[jgraham]

(I am not an expert, etc.)

The only transition path in terms of changing licences would be to adopt the only the BSD variant currently allowed. Relicensing wpt is effectively impossible, since there is no copyright assignment and there are such a large number of contributors.

I'm all in favour of dropping the W3C Test License part though; I don't think anyone uses that in practice.

[foolip]

There is also CONTRIBUTING.md, which I do not fully understand. @plehegar, can you explain in lay terms what that means, and how it relates to the license, if at all?

[plehegar]

The first is the grant of license, which basically gives us the right to relicense under the original BSD license, effectively replacing the current LICENSE.md if we'd like to. I could check to see if we still need to have it but I would guess so, except that we could give the right to relicense to more than just W3C.
The second part is a disclaimer. not sure if we need this.

[foolip]

@plehegar, by "the original BSD license", do you mean something other than the "W3C 3-clause BSD License" in LICENSE.md?

[plehegar]

well, in CONTRIBUTING, we say "under a BSD License". So, it's not clear to me that we're restricted to the W3C BSD one.

[wseltzer]

It's not quite that dire, @plehegar :) we added the license.md text here in 2016.

That said, if there are practical changes that would help people in making or using tests, I'm all ears to help.

[plehegar]

A bit more history for background.

The license in the repo actually dates back to December 2012:
4777e4c#diff-04c6e90faac2675aa89e2176d2eec7d8

The grant of license is from 2013.

Prior to December 2012, the contributions were inherited from W3C Groups, thus under the dual license dating back to July 2008 or so:
241b48c

Prior to 2008, we were using the W3C software license. I don't know how much of that original test suite (Early draft of the CSS 2.1 test suite, inherited from the CSS1 test suite) survived since we didn't have ref tests at the time.

The current LICENSE today has this dual aspect which I'm no longer sure makes sense. Thus I'm wondering if we could at least get away with just the 3-clause BSD license nowadays.

[foolip]

@wseltzer, could you comment on whether reducing LICENSE.md to just the second license in there is doable?

[wseltzer]

@foolip yes, I believe we could switch to just the 3-clause BSD license instead of the current dual-license. (Anyone who was using tests under the W3C Test Suite License would be able to continue to do so.)

[foolip]

@wseltzer, that sounds great, do you want to prepare a PR to trim LICENSE.md so that we get it exactly right?

[annevk]

I looked into providing the PR for this, taking https://opensource.org/licenses/BSD-3-Clause as inspiration. However, that suggests you need to start with

Copyright <YEAR> <COPYRIGHT HOLDER>

which is something that we currently lack. Would

Copyright 2018 web-platform-tests contributors

work?

[foolip]

I think the idea is to trim LICENSE.md to just contain https://github.com/w3c/web-platform-tests/blob/master/LICENSE.md#w3c-3-clause-bsd-license verbatim, but waiting for PR from @wseltzer or @plehegar instead of trying it myself :)

[annevk]

Sure, but that wouldn't comply with the official license format as far as I can tell. It also seems strange to preserve the W3C prefix.

[foolip]

@annevk do you mean a format that GitHub expects?

[annevk]

That the license, as per https://opensource.org/licenses/BSD-3-Clause, should be in.

[foolip]

There are a bunch of other minor differences too, like "source code" vs. "works" and "COPYRIGHT OWNER" vs. "COPYRIGHT HOLDER".

The label of the license is presumably not part of the license and could be dropped, but this bit is part of the license itself: "Neither the name of the W3C nor the names of its contributors may be used to endorse or promote products derived from this work without specific prior written permission."

@plehegar @wseltzer, do you believe any change to that is required/desired?

[foolip]

Ping @plehegar @wseltzer, any progress on this?

[wseltzer]

I'm sorry, but PSIG (the Patents and Standards Interest Group, which gathers IP lawyers) wants to discuss further, and proposes to do so at their July 9 meeting. I have recommended that we use the second of the dual licenses -- the W3C 3-clause BSD, because we're not aware of any continuing (or past) need for the protection against claims of conformance, but I think we owe them the opportunity to discuss and raise concerns. If there's other input from here to that discussion, I'll share that then, Thanks.

[foolip]

Thanks @wseltzer, please keep us posted! The question in #11009 (comment) and whether "W3C nor the names of its contributors" can be replaced by "web-platform-tests project nor the names of its contributors" are also questions that I'd like to see addressed.

[annevk]

@wseltzer see #11191 (comment). In particular:

[T]hat variant doesn't have the copyright line and refers to the W3C twice, which seems weird if this is meant to be BSD. Are those considered identical enough for projects to use it without issue?

The lack of a copyright line seems pretty problematic, not sure about wording differences.

(On reflection, this seems identical to @foolip's comment.)

[wseltzer]

Per conversation with @foolip I'm filing two pull requests against license.md, one dropping from the dual license to the W3C 3-clause BSD; the other replacing with the straight 3-clause BSD (which differs only in not having the named copyright holder W3C). PSIG has given non-objection to the W3C 3-clause BSD. If that would still cause legal hiccups, then we can probably persuade them of the vanilla 3-clause.

[jugglinmike]

We may be able to resolve these issues using WPT's new RFC process:

web-platform-tests/rfcs#18

[jayvdb]

I note that the PyPI source distributions do not contain a LICENSE file (at least two I checked). It is highly desirable that they are included, usually by adding it to MANIFEST.in.

[foolip]

Finally resolved via web-platform-tests/rfcs#18 and #11191, thanks everyone!