I have a question regarding licensing. There is one question that asks essentially the same thing, however, I'm not satisfied with the answers given. Can you help me refining my questions so that I can ask it and get all my points addressed without posting a duplicate?
This is existing question: License violation within git history . My situation is very similar, except that I neglected to put license headers in the first revisions and I'm wondering if these older revisions will essentially be unlicensed once I set my project to public.
All answers in the existing question hinge on manipulating the code base (by leaving out files or manipulating the blob). That will always be possible and has no bearing on my question: In these cases the author published a correct version and another person manipulated it. In my scenario, the bona-fide as-published version does not have a license (or violates the license in the case of the existing question).
I think the core question is: Does an old (not-tagged) revision count as a stand-alone publication which needs a license or can a user be required to take an actual release into account? The ground rule is: A published version needs a license in every source file or it is not licensed. Does a revision in the VCS count as publication in that sense or only revisions tagged for release?