Re: What is the "open web" ?

On Wed, Jun 5, 2013 at 12:39 PM, Jeff Jaffe <jeff@w3.org> wrote:
> On 6/5/2013 5:08 AM, Henri Sivonen wrote:

>> People who say production CDMs are incompatible with Open Source
>> observe that they can't come with the downstream freedoms associated
>> with Open Source. I.e. they consider what freedoms the recipients of a
>> production CDM receive. You seem to be looking at this from the
>> perspective "If a CDM vendor receives code under Open Source license
>> Foo, can the CDM vendor incorporate that code into a CDM without
>> violating the license the code came under?" Obviously, the answer is
>> "yes" for Open Source licenses that don't try to ensure that
>> intermediaries don't reduce downstream freedoms.
>>
> Exactly.
...
> I had asserted that EME was implementable in open source from the
> perspective of the license.

In the sense of code going into a CDM implementation being under an
Open Source license, yes. In the sense of the CDM build that's
distributed for deployment meeting the Open Source Definition, no.

It's the latter that matters. As a more traditional example, consider
OS X. A lot of Open Source code goes into OS X. Yet, OS X as
distributed is not Open Source as a whole.

Also, not violating the terms of an Open Source license is alone
pretty worthless in terms of determining if a binary given to you
meets the Open Source Definition. For example, the MIT license is a
certified Open Source license, but not violating the terms of the MIT
license is not sufficient to comply with the Open Source Definition.
As a simple example, the MIT license doesn't guarantee the inclusion
of source code, even though the Open Source Definition requires
including source code.

> Norbert asserted that if a spec is not implementable in GPL [3] it is
> probably not implementable in open source.
>
> I think we all recognize that this is not accurate, that a spec can be
> implementable in open source but not GPL.

If something cannot be implemented without violating GPLv3, it's a
pretty good predictor that an implementation can't be shipped without
failing to meet the Open Source Definition *for what's shipped* (as
opposed to the ingredients of the implementation).

> Which does not mean that it is a good idea or provides downstream freedom.
> That indeed is a different debate.

No. "Can downstream freedom be provided" is at the core of the "can be
implemented as Open Source" question. It's not about "could it have
been Open Source prior to the point of removal of downstream freedom".

-- 
Henri Sivonen
hsivonen@hsivonen.fi
http://hsivonen.iki.fi/

Received on Wednesday, 5 June 2013 12:27:18 UTC