licensing review: drop non-commercial and no-derivates terms?

dkg commented 1 year ago

Over in https://bugs.debian.org/1028570, i observed that the licensing for these files is not compatible with the debian free software guidelines (DFSG).

Is it possible to relicense it from CC 4.0 BY-NC-ND to just CC 4.0 BY? or CC 4.0 BY-SA?

I contributed 015-arabic/* and I'm certainly fine with those files being relicensed in such a way. I'm not sure of the origins of the other files in question.

Without relicensing, it's going to be a challenge to ensure that the package's tests run correctly in Debian, since all build and test infrastructure in the project is intended to use only DFSG-free material.

MartinThoma commented 1 year ago

The van run the tests without the sample files.

I have to think about changing the license. It might be that only the two of us have contributed so far.

dkg commented 1 year ago

Not sure what you mean by "the van run the tests" -- can you elaborate? Are you saying i should just run the tests without the sample files? I assume there will be less coverage in that case.

If it's possible to change the license, that would simplify matters a lot for me. Thanks for considering it!

dkg commented 1 year ago

fwiw, the scope of coverage is already limited because debian test suites don't use the internet at all , and we have a patch to disable the network-reliant tests. The more the tests can be done on local copies of files that share similar licensing terms to the software itself, the more likely we'll be able to have the full test suite run on all the different architectures that debian's testing infrastructure supports.

MartinThoma commented 1 year ago

I don't see the point of running the tests on different architectures as this is all about Python. If (and only if) it works for any architecture with a given Python version, it will always work for all architectures with that Python version.

MartinThoma commented 1 year ago

Or in other words: pypdf runs its tests on Python 3.6 - Python 3.11 (see GitHub CI) and officially supports 3.6 - 3.11 (see trove classifiers).

The main thing that might be interesting to test is stuff that requires additional packages (pycryptodome / Pillow) to check if the Debian packaging broke anything :thinking: I'm not sure how that actually works.

MartinThoma commented 1 year ago

I hope I find the time at the end of the day to go over all files / make up my mind if this change of license is possible / a good idea.

The license you propose would be more permissive, but people could sell the stuff in there without having ever contributed, right?

dkg commented 1 year ago

Yes, with standard CC BY licensing it would be completely permissive. with CC BY-SA it would be closer to a copyleft situation: anyone transmitting the file to anyone else would need to "share alike".

dkg commented 1 year ago

Any progress on this? In debian, i've moved ahead by just dropping tests on these sample files, but if they get a DFSG-free license, i'd be happy to ship them as a corpus and the nexpand the pypdf test suite; or we could include them as a "second tarball" for the source package for pypdf directly.

MartinThoma commented 1 year ago

Sorry, I forgot this one. Let's go through it: