Project: choose a license

balbasty / nitorch

Neuroimaging in PyTorch

Other

83 stars 14 forks source link

Project: choose a license #8

Open balbasty opened 4 years ago

balbasty commented 4 years ago

We need to decide on a license before going public. I'd rather be very permissive.

A solution is GNU-GPL, which is quite permissive but forbids users to distribute closed-source software that would rely on nitorch. That would probably be a big blocking point for you @brudfors
Another solution is the MIT license, which lets people do whatever they want, including develop closed-source software. The issue then comes from the spm module: are we limited by the SPM license, which is GNU-GPL? (I think we are). We'd need to discuss this with JA and GF.

For info: https://choosealicense.com

brudfors commented 4 years ago

How does it work if you re-implement code in a different language? Does the original code's licence transfer?

What is the perk of being permissive?

balbasty commented 4 years ago

If the structure of the code is the same, it seems that the license transfers: https://opensource.stackexchange.com/questions/6699/if-i-translate-port-a-repository-from-one-programming-language-to-another-do-i

brudfors commented 4 years ago

Fair enough, why not use MIT then? If you are happy with that?

brudfors commented 4 years ago

Maybe this makes it more suitable to have a nitorch-tools, with a different license?

balbasty commented 4 years ago

Yes, we could have nitorch-spm with a GPL license that contains pure ports. I think I'd like to have MIT for the core project, because it makes it more widely usable. And I think it would also be potentially beneficial for your project (which I assume will be closed-source if commercial). Does PN know that SPM is GPL?

balbasty commented 4 years ago

PyTorch is (I think) BSD, which is basically MIT + credits required. But we don't compile or package anything from PyTorch ourselves, so I don't think it makes a difference.