Re: Spec license

On Feb 3, 2009, at 17:32, Karl Dubost wrote:

> Le 3 févr. 2009 à 10:10, James Graham a écrit :
>> I should be able to copy sections of the spec text into the source  
>> code or testcases and license the whole under an MIT license (I  
>> don't recall how often we actually do this in html5lib but I  
>> certainly would like to do so more often in the future).
>
> I have difficulties to understand why it is not possible with the  
> current W3C Document License. What you are doing is not a derivative  
> work nor a full copy of the document, but just quotations of it,  
> which is already authorized by the document license.

Relying on the American Fair Use doctrine or European enumerated but  
subjective limitations[1] of copyright such as the right to quote are  
not good enough for inclusion of content in a piece of software that  
is then as a whole explicitly licensed under an Open Source license.  
For an Open Source software whose project management is itself  
properly diligent and wants to get the software included in other  
projects whose management is diligent, including substantial passages  
from a spec under a subjective limitation of copyright (where  
subjectivity is ultimately up to a judge) without an explicit license  
is not acceptable.

In case the above paragraph seems paranoid, here's where I'm coming  
from: Previously a member of W3C staff has relayed to me a "warn[ing]"  
from W3C lawyers because I had merged the prose of the HTML5  
tokenization section into the source code of the Validator.nu HTML  
Parser as comments. I had copied the text from the WHATWG copy of the  
spec, which permits such copying. The "warn[ing]" was quickly  
retracted, but I think it is enough to show that it would be unwise  
for a software developer to undertake copyright-sensitive actions on  
the text of an HTML WG deliverable unless the action is permitted by  
an explicit license.

In short, if we wish to address a use case, the use case needs to be  
permitted by the spec license. Saying that it counts as a quote or as  
Fair Use is not good enough.

> Maybe we should first identify what are the use cases and see if the  
> set of licenses, we have from W3C Document Licenses to others,  
> covers or not the use cases.
>
> So far I see
>
> * Publishing the full or parts of a specification in a book to be  
> sold.
> * Include prose of the specification in software from proprietary to  
> complete open source
>
>
> Something else?

  * Copying the prose defining an algorithm, pasting it verbatim or  
with modifications into source code of a program as comments and  
writing an implementation of the algorithm (possibly making creative  
optimizations) so that the spec text and the statements of the  
programming language intermingle. The resulting program should be  
licensable under the MIT license without additional terms. (For  
algorithms contained in "HTML 5", this use case is addressed by the  
WHATWG license, but relying on that means that the W3C instance of  
HTML5 can't be treated as the canonical instance. I'm already  
exercising this option on the source code of the Validator.nu HTML  
Parser under the WHATWG spec license.)

  * Extracting parts of a spec and showing the extracts verbatim or  
with modifications in the user interface of a validator. The validator  
as a whole including its UI strings should be Open Source and should  
be suitable for packaging in popular GNU system distributions  
including Debian. Both distributing a copy of the spec text and a  
program for extracting pieces of it at runtime and distributing  
preprocessed extracts should be permitted. Showing a copyright notice  
on in the documentation of the validator is acceptable, but showing a  
copyright notice or other legal legends in the UI whenever a spec  
extract appears would not be acceptable. (I'm already doing this in  
Validator.nu under the WHATWG spec license. However, I'd be interested  
in having the option to do this with the authoring guide/reference  
deliverables of the WG and with the SVG and MathML specs.)

  * Extracting the WebIDL parts of the spec and incorporating them  
verbatim or with modifications into the source code trees of  
implementations of the spec. Such inclusion should not interfere with  
LGPLv2.1 or later (Gecko, WebKit), GPLv2 or later (Gecko), MPL 1.1 or  
later (Gecko), Apache Software License 2.0 (Batik) or the licensing  
models of Opera and IE.

  * Extracting the CSS parts of the spec and incorporating them  
verbatim or with modifications into the source code trees of  
implementations of the spec. Such inclusion should not interfere with  
LGPLv2.1 or later (Gecko, WebKit), GPLv2 or later (Gecko), MPL 1.1 or  
later (Gecko), Apache Software License 2.0 (Batik) or the licensing  
models of Opera and IE.

  * Copying prose from the spec and pasting it verbatim or with  
modifications into comments or accompanying documentation of a test  
case and checking the test case into the source tree of any of the  
above-mentioned software projects plus html5lib without interfering  
with their licensing or choice of project hosting.

  * Continuing the development of the WG deliverables in a non-W3C  
venue if the W3C or the HTML WG cease operations.

  * Forking some or all of the WG deliverables and pursuing an  
alternative development path outside the W3C even without the W3C or  
the HTML WG ceasing operations. (That is, I think what Rob Burns is  
doing on the HTML4All wiki should be allowed.)

I believe licensing the spec under the MIT license[2] would address  
all the above use cases to my satisfaction.

Additional use cases that I think are worthwhile but that I'm less  
interested in advocating personally:

  * Taking WG deliverables in whole or part and repurposing content  
into a book that is given gratis or sold on paper or as a digital file.

  * Using extracts from WG deliverables in documentation published on  
developer.mozilla.org.

[1] http://en.wikipedia.org/wiki/Limitations_and_exceptions_to_copyright
[2] http://www.opensource.org/licenses/mit-license.php
-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Wednesday, 4 February 2009 08:43:32 UTC