Thomas Klausner <wiz%NetBSD.org@localhost> writes: > What do you suggest as criteria to decide if a random text is near > enough to a BSD license? Especially with kind-of-BSD licenses I find > it very hard to tell. Basicially you have to diff the text and apply acceptable wording fuzz judgement. Licenses are legal documents so they really should be sort of close. But, the point of pkgsrc license tagging for Free/Open Source licenses is not entirely clear to me. The only use case I can think of is someone who objects to some license that has unusual terms such as trademark claims, or is too complicated to understand (apache 2?), or is copyleft when they want BSD. I can't think of a reason for someone to object to a BSD-like license when they broadly do not object to BSD-like licenses. So, we could A) have a license file "original-bsd-like" that says that the license is similar in spirit to the 3-clause BSD license, but not identical, and those that don't like this can 1) take original-bsd-like out of DEFAULT_ACCEPTABLE_LICENSES 2) get FSF and OSI to rule on the license in question 3) send a patch to add the offending license as a file with or without -license depending on 2, and add it to DEFAULT_ACCEPTABLE_LICENSES also depending on 2 I think we will find that no one cares enough to do the steps B) Do the steps above. But I think no one cares enough. I am in favor of encoding that we are being sloppy, rather than pretending things are the same when they aren't. The rule would be that if anyone makes a credible objection on tech-pkg that a package tagged as original-bsd-like has a license that is not substantially equivalent, then we retag the package to generic-nonlicense until someone sends a patch to add the actual license (again with steps 1-3).
Description: PGP signature