Re: libpng license 2.0

Bob Friesenhahn wrote:
> There have been many contributors to libpng, and any of them which
> made a significant contribution are copyright holders with the
> expectation that their contribution is under the license which was
> current at the time the contribution was made.

Yes, and they're all listed in the present-day libpng license under
the terms and conditions of that time, and that's still unchanged in
my proposal at

https://gist.github.com/ctruta/5e276eb83213f9d66bf61539156830c5

See the "PNG Reference Library License version 1" subsection, which is
part of the "COPYRIGHT NOTICE, DISCLAIMER, and LICENSE" section.

> Regardless of a modified license statement, the previously existing
> licenses are perpetually still in effect unless all the significant
> contributors agree to the new license.

Correct. I'm complying to that already, I think.

On a second look, this is "almost" correct. Comparing the licenses of
libpng-1.6.34 vs libpng-1.0.7, I saw statements inserted after the
fact, like, for example, this one:

  "Some files in the "scripts" directory have other copyright owners,
   but are released under this license."

inserted recently (2016) by Glenn, in the middle of a statement dated
1996-1997 and copyrighted by Andreas Dilger. I don't mean to say that
it was legally correct, or not, to insert that statement. Only to point
out that it had been done.

> Usage of the software then
> needs to comply with both the old and new license statements.

That would be an issue if the two licenses had different terms and
conditions. But the zlib license and the libpng license have the same,
or at least very similar terms and conditions, and this is why OSI
approved only one of them (zlib). One thing that OSI doesn't like is
redundant licenses.

This is the story that I've been told. I wasn't around at the time.
My knowledge may be incorrect, and I will appreciate corrections.

> It would be nice if the libpng license and copyright notices were
> cleaner but the reality is that the existing statements are appended
> to the new one (only applying to new work) rather than the new license
> replacing the old one since you don't have the legal authority to
> replace it.

You meant to say prepended? (Yes I'm nitpicking, but this is licensing,
and we must aim to be 100% precise.)

I'm prepending one more ("libpng license 2.0"), in the hope that this
will be one last licensing update, and anything that follows shall be
new one-liner copyright notices and copyright years only.

I wish I didn't have to do this. I would have gladly submitted the
license that came with libpng 1.6.34, verbatim, to OSI, for "legacy
approval", then close that book. But I can't, because of the wording
that makes it almost obligatory to continue maintaining the license,
as libpng itself is being maintained.

Glenn didn't mind maintaining the license, and updating it at every
libpng release, and at every newly-acknowledged contributing author.
But I'd rather not continue doing that.

> Take care not to muddy things up further while trying to tidy up the
> license.

Thanks. I want to do this once, submit it to OSI for formal review,
incorporate their recommendations under the assumption that they
actually know how this works, and then continue focusing on what
truly matters: hacking libpng.

There have been precedents with newer licenses superseding older ones.
It's been rarely done, but it is possible, and the Mozilla Public
License going from MPL-1.0 to MPL-1.1 to MPL-2.0 is a case in point.

https://opensource.org/licenses/do-not-use.html

Can the "libpng license 2.0" supersede the current "libpng license"?
Flatten the list of contributing authors (as it should have been done
from the very beginning, in my opinion), and settle for good at a
non-changing license text?

Or, rather, can the new text only come in addition to the existing one
(as in my proposal), and settle at a no-longer-changing license text?

Or, (another alternative that I would be ok with), can the current one
be modified in a way that will not require future maintenance?

I'm glad that libpng-1.6.35 looks stable enough, without any serious
defect that would require an emergency 1.6.36 release.

The license review process at OSI is said to take around 30-60 days.
They recommend holding off publishing any new releases until they are
done with their review, to be sure that their recommendations do get
incorporated.

I think (and hope) we can do that.

I will appreciate any input in this discussion, while I prepare the
formal submission to OSI.


Marcus Johnson wrote:
> Why not just move the license over to a standard one? like MIT, BSD, Apache, etc

Relicensing without explicit approval of all copyright holders is not
possible in general. It is possible only in a few exceptional cases.
For example:
* From GPLv2 to GPLv3 and from LGPLv2.1 to LGPLv3
* From LGPLv2.1 to GPLv2 and from LGPLv3 to GPLv3
* From 2-clause BSD to 3-clause BSD
* From EUPL to GPL/LGPL/MPL/EPL/CPL/CeCILL

And not much else.


Sincerely,
Cosmin

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot