Re: CVS commit: pkgsrc/licenses

[Ryo ONODERA <ryo%tetera.org@localhost>]

Hi,

Sorry. I have replied to pkgsrc-changes@.

Greg Troxel <gdt%lexort.com@localhost> writes:

> [moved to tech-pkg; see
>
> https://mail-index.netbsd.org/pkgsrc-changes/2021/02/20/msg229535.html
> https://mail-index.netbsd.org/pkgsrc-changes/2021/02/21/msg229543.html
> https://mail-index.netbsd.org/pkgsrc-changes/2021/02/22/msg229633.html
> ]
>
> Joerg Sonnenberger <joerg%bec.de@localhost> writes:
>
>>> However, I found that the FSF considers this to be a Free license:
>>> 
>>>   https://www.gnu.org/licenses/license-list.html#ODbl
>>> 
>>> Thus, I think this should be renamed to odbl-v1 and put in
>>> DEFAULT_ACCEPTABLE.  I am guessing this is not controversial.
>>
>> Are you sure? The text is very long and I'm not sure how similar it is to
>> the AGPL. Language like the termination of the license certainly don't
>> sit well...
>
> Many Free Software license have termination conditions, such as GPL2
> (section 4), so I don't see that as unusual or particularly concerning.
>
> Separately from reading the text, I am familiar with this license within
> the OpenStreetMap community.  OSM used to use cc-by-sa, but it turns out
> that database copyright is complicated legally because different
> jurisdictions have different notions about databases.  In particular the
> US does not recognize database rights, and Europe does.  That caused
> cc-by-sa to not really work in Europe for databases.  AIUI, the ODbL was
> written so that it would be valid in the various jurisdictions.
>
> I can understand your concern about similar-to-AGPL, but I don't think
> this license is like that.  It defines Convey:
>
>   "Convey" - As a verb, means Using the Database, a Derivative Database,
>   or the Database as part of a Collective Database in any way that enables
>   a Person to make or receive copies of the Database or a Derivative
>   Database.  Conveying does not include interaction with a user through a
>   computer network, or creating and Using a Produced Work, where no
>   transfer of a copy of the Database or a Derivative Database occurs.
>
> So that means that rendering a map based on an ODbL-licensed database,
> as you would see at https://www.openstreetmap.org/, is not "Conveying".
>
> As an aside for context about AGPL, I think there are multiple things
> going on that lead to concerns for pkgsrc and the resulting "AGPL may
> not be in DEFAULT_ACCEPTABLE even though it is Free".
>
>   Some people within the pkgsrc community are offended by AGPL and don't
>   like it.
>
>   The notion that someone could install something and configure a web
>   server and have an obligation to distribute the unmodified version
>   they got.  This is not that big a deal (because it's already
>   available) and I have not heard of anyone being hassled about this.
>   And, most AGPL web services are set up for a "source download" button.
>
>   The notion that someone could get something from pkgsrc, spend effort
>   to make modifications that they want to keep secret and then use it as
>   a web service because they don't understand because they relied on
>   pkgsrc's license labeling.  I think this is a strawman and haven't
>   heard of a case.  I'm also unsympathetic, as anyone trying to make
>   proprietary modifications to Free Software needs to understand what
>   they are doing and I very strongly believe that it is not pkgsrc's job
>   to do this for people.
>
>   The situation where there are two distinct usages of AGPL: 1) projects
>   that are entirely Free Sofware that wish to prevent proprietary
>   derived works being used outside of organization, and 2) "projects"
>   that engage in "proprietary dual licensing" where the primary intent
>   is to sell proprietary licenses.  Type 2 projects tend to be
>   heavyhanded about AGPL terms to sell licenses.  Broadly, the hard-core
>   Free Software world views Type 1 as legitimate and Type 2 as not.
>   Overall, if there were only Type 1 uses, I think that we might not
>   have ended up where we did.
>
> I am not aware of anything licensed under ODbL that is Type 2.
>
> The license is long, but I think that's because it's modern and aimed at
> some complex provisions of EU law, not only database rights but also
> moral rights, while also trying to function in the US and other
> jurisdictions.
>
> There is a share-alike provision, but I see it as GPL-like, not
> AGPL-like.  Suppose you modify the database, and then make a map from
> it, and then publish that map.  You are required to provide the modified
> database.  This is like modifying source code and creating a binary and
> putting that binary on your web server; the database is like source and
> the rendered map is like a binary (it is not editable, and it is the
> form that users use).  The AGPL, in contrast, triggers the requirement
> to distribute if you let someone interact with the program but do not
> transfer a binary to them.
>
> In the OSM world, in terms of the proprietary dual licensing concern OSM
> is firmly Type 1, I have heard zero of any Type 2 enforcement/agitation.
> All of the angst -- which is considerable -- is about people that
> produce maps from OSM data and fail to provide the required attribution.
> This is akin to compiling source code and removing copyright notices.
>
> You did raise a good question, but after digging into it more I conclude
> that ODbL V1 belongs in DEFAULT_ACCEPTABLE under our current policies.
>
> I am curious what's to be imported that's going to use it...

My answer to the last question is as follows.
From my post to pkgsrc-changes@:

> At first, ODbl 1.0 was added for upcoming inputmethod/rime-cantonese.
> However I have found that rime-cantonese and fcitx5-rime have
> technical problem just before commit and I will not add rime-cantonese
> for a while longer.

-- 
Ryo ONODERA // ryo%tetera.org@localhost
PGP fingerprint = 82A2 DC91 76E0 A10A 8ABB  FD1B F404 27FA C7D1 15F3