On Fri, Oct 13, 2017 at 02:30:18PM -0400, Richard Fontana wrote:

By the bye, one thing I'd find useful, either inside or outside of
SPDX, is some notion of correspondence of an FSF-approved license
with a counterpart OSI-approved, or SPDX-recognized, license.

To illustrate, consider the MIT license. There is no MIT license
steward; the de facto standard text (basically for historical
reasons) is that on the OSI website,
https://opensource.org/licenses/MIT

I agree, and think the best way to maintain that sort of information
is with automated SPDX license matching [1]. Then, everyone with
license opinions (the OSI, FSF, etc.) publishes a list of reviewed
license IDs (using SPDX IDs or their own custom IDs, e.g. [2]):

$ curl -s https://api.opensource.org/licenses/ | jq '[.[] | .id]'
[
"AAL",
"AFL-3.0",
"AGPL-3.0",
…
]

Then for each reviewed license, they provide a way to get the
metadata. For example [3]:

$ curl -s https://api.opensource.org/license/AAL | jq .
{
"id": "AAL",
"identifiers": [
{
"identifier": "AAL",
"scheme": "SPDX"
},
…
],
…
"text": [
{
"media_type": "text/html",
"title": "HTML",
"url": "https://opensource.org/licenses/AAL"
}
]
}

And the text. The OSI doesn't have a good example of this at the
moment; you'd have to scrape the wrapping markup off of [4].

Then you can use regular SPDX matching to determine the SPDX ID for
the license in question to determine an objective (and verifiable)
match (modulo wiggle room in the SPDX matching definition ;).

Individual providers may suggest their own ID mappings (as in the OSI
example above, where the OSI asserts their AAL license has the ‘AAL’
SPDX ID), and downstream consumers could either trust those mapping
assertions, verify the asserted mapping, or come up with their own
mapping.

The FSF speaks of a license it calls the X11 license as being a free
software license and says that this is sometimes called the MIT
license (a label the FSF has long considered confusing or
misleading). However, the license pointed to by the FSF is not
textually identical to the OSI MIT license and also does not match the
SPDX license "MIT", but does match the SPDX license "X11".

They also list the Expat license as free and GPL-compatible [5], and
it matches the SPDX MIT [6]. So you can say the FSF considers the
SPDX MIT free and GPL-compatible.

On Fri, Oct 13, 2017 at 01:09:43PM -0600, J Lovejoy wrote:

On Oct 13, 2017, at 12:02 PM, W. Trevor King wrote:

On Fri, Oct 13, 2017 at 10:20:33AM -0700, Gary O'Neall wrote:

There is a request by the FSF and approved by the legal team to
add a property to the listed licenses isFsfFree to indicate if a
license is identified by the Free Software Foundation as a Free /
Libre license. This would be a simple Boolean type.

I am against this in license-list-XML, for the same reasons I am
against our current osi-approved type: SPDX should not be a
canonical source of whether *someone else* has approved a license
or not. I'd much rather provide tools for Alice to start with an
SDPX ID and lookup Bob's notes for a given license. More on this
in [1].

This is not really a for or against issue :)

It is something that has been asked for by the SPDX community (David
Wheeler, last person who asked) and discussed as something that
would be nice to have, but we’d need help from the FSF to organize
and maintain. Now that the FSF has asked for this directly, we are
adding it. SPDX is not claiming to be a canonical source here, just
reflect the same info found elsewhere. Given anyone will be able to
submit a pull request, the expectation once the initial work is
done, will be that the FSF will help keep this current if there are
changes.

Right. I'm not calling for “never document this anywhere”. I'm
calling for only documenting the OSI ID, FSF ID, etc. for each license
in license-list-XML and maintaining those mappings with automated
tooling. Then downstream resources built from license-list-XML (like
license-list-data [7]) can bake in whatever OSI-provided and
FSF-provided metadata they want for their consumer's convenience
without worrying about maintaining it in license-list-XML.

As for the OSI information - this was decided to be included very
very early on (probably 2010, or 2011 - SPDX License List got going
in fall of 2010, if memory serves) and again as a joint decision
with the OSI. At that time, they did not have the API.

Supporting existing users is important, which is why the old fields
should stay in license-list-data. But nobody's relying on
license-list-XML (yet), so we aren't constrained by backwards compat
there. As long as we can recover the old information when compiling
license-list-data (and I think we can), I don't see a backwards-compat
issue.

if there is some improved way to get to the same end point (of
identifying on the SPDX license list, which licenses have been
OSI-approved), then we can look into that, but I’d think it’d take
some work from both SPDX and OSI to do so.

I've been trying to provide some of that work ;).

This sounds like a nice idea… but we have to work with what we have
and FSF does not have such an API…

But now they're asking for one. Can't we help them build one, instead
of helping them stuff their metadata into our API? It seems like
better separation of concerns that way.

Cheers,
Trevor

[1]: https://github.com/spdx/license-list-XML/issues/418
[2]: https://github.com/OpenSourceOrg/api/blob/master/doc/endpoints.md#licenses
[3]: https://github.com/OpenSourceOrg/api/blob/master/doc/endpoints.md#licenseid
[4]: https://opensource.org/licenses/AAL
[5]: https://www.gnu.org/licenses/license-list.html#Expat
[6]: https://directory.fsf.org/wiki/License:Expat
[7]: https://github.com/spdx/license-list-data

--
This email may be signed or encrypted with GnuPG (http://www.gnupg.org).
For more information, see http://en.wikipedia.org/wiki/Pretty_Good_Privacy

W. Trevor King wrote:

They also list the Expat license as free and GPL-compatible [5], and
it matches the SPDX MIT [6]. So you can say the FSF considers the
SPDX MIT free and GPL-compatible.

Ah right - so not as interesting an example as others I was thinking
of. I've been thinking about this because there's been some interest
among the FSF and OSI in seeing where exactly the lists of
FSF-recognized-as-free and OSI-approved licenses disagree.

It's an accident of history that the Expat license text, rather than
the X11 license text, is what came to be known as the MIT license, I
suppose.

Richard