I have previously written a lot of code that is hosted on a public repo on GitHub, but it never had a license. It was written as part of my work while working for a non-commercial academic entity, and I would like to add a license before the link to the repo will be included in something that will be made public, potentially attracting one or two visitors.
This leaves me with a couple of questions:
- Can I just add a license after the fact and it will be valid for all prior work?
- Do I have to make sure the license is included in all branches of the repo, or does this not matter? There are for instance a couple of branches that are used to freeze the state of code at a certain time for reproducibility’s sake (I know this could be solved in a better way, but that’s how it is).
- I have myself reused some of the code in my current work for a commercial entity (internal analysis work, only distributed within the organization). Should this influence the type of license I choose? I am considering a GPL-license, but should I go with (what I believe to be) a more permissive license like MIT because of this?
Thanks for your answer.
OK, in that case it may not even make much sense to add a license. There will be no added code to this repo in the future, so there will nothing the new license would apply to.
Yes, you kind of answered this in question 1. Since it is not retroactively applied, it won’t apply to the stale branches that only exist as snapshots of the code.
Good point. This is not included in any software that is distributed, it is only a smaller part of an internal codebase used for data analysis. Does that not change things? But to be on the safe side, it would probably make sense to make it as permissive as possible to avoid any issues here. But then again, if it is not applied retroactively then nothing of the code used will be subject to any license. But good thing to remember for the future.
Either you misunderstand or the person you are responding to is. If you retroactively add a license to the current state of the code (for example by committing a new LICENSE file and adding the new license to the top of each file), or course that applies to the entire state of that code as of that commit. What is more difficult is that earlier commits won’t have that license explicitly unless you rewrite git history to make that happen (which is possible but tedious).
You can always relicense code you own the rights to. You can even dual license it, or continue to use it commercially in terms contradicting the license you open sourced it as, as long as you have the permission of every contributor.
The idea that a license added would only apply to code added after the license change is very funny.
Yeah to be clear, I wasn’t suggesting you can’t change the license at all, or it requires code changes or whatever.
You are completely free to apply whatever license, and use that going forward without changing the code at all.
And like you have correctly pointed out, you could rewrite git history, or even just remove all prior versions of the code.
I am of course happy to be wrong or have misunderstood something- I am absolutely not an expert and would like to be corrected if I am wrong!
I suppose it makes sense if it originally had a license, and you then change the license to be less permissive.
I will probably not do that, but I guess it factors into my second question: That I in that case should make sure to include it in all branches (which are not treated as branches in the common sense, but rather as forks within the repo - they will never be merged to the main branch).
Or just write an external document saying that all files in all previous revisions are available under whatever licence. There is nothing magical about a licence notice at the top of a file. It is just helpful documentation of the license. But you can document a licence anywhere as long as you are the owner of that code.
Also not a lawyer, but you can also grant exceptions to the license (if you’re the sole owner of the code), so you can license code one way and let a certain org use it another way.
Which is essentially already what’s happening. The default “license” of something is that you have full ownership and no rights are given to anyone else. You’ve essentially give your company an exception to use it for that project.