Dec 21 2019

Licensing software to be permissive

project.

A while back we were discussing changing the license of transonic from CeCILL-B to a more recognizable, but still permissive license. We would soon make that move. The most obvious candidate was the 3-clause BSD license and I suggested Apache 2.0 as an alternative, due to the potential benefits.

Then I saw this:

Being a licensing nerd, I found this amusing piece informative on why everyone should avoid MIT and BSD. Patents and ambiguity. The article also links to verbose explanation of what the MIT license actually means, along with this warning:

As a result, the shortness of MIT and BSD are reassuring only until you actually try to understand them, and find you need a decoder ring. The terms are actually dangerous if you read them without knowing you need that decoder ring, believing you see the whole picture, which often turns out to be just the picture you wanted to see.

A friend of mine had used the MIT license a while ago, commenting that he understood every word of it! Well, did he?!

My arguments in favour of Apache 2.0 were as follows:

1. I don't think Apache 2.0 makes any demand from users of transonic or vice-versa. It is compatible with BSD too

2. To quote the license itself from section 1, it makes it adequately clear what derivative works are:

For the purposes of this License, Derivative Works shall not include works that remain separable from, or merely link (or bind by name) to the interfaces of, the Work and Derivative Works thereof.

Source

1. The key aspect of Apache 2.0 is it protects developers from patent lawsuits

Indeed something more easier to read, but legally sound would be nice, like Blue Oak.

Note: Don't use the Blue Oak license ... yet. Once it is approved / vetted by GNU and OSI we can start using it. I seems like a good license, but I am no expert, and most likely you are not too. Unfamiliar licenses slows down contributions and use of software. It is listed on SPDX though!