License R code when using R packages with different license types

Question

I am navigating licensing, and this is very confusing.

I produced a lot of R code to analyze data for a research article that will be published on a peer-reviewed journal.

Reading here License for R scripts, I can understand that I would need to pay attention to the license type of the packages I used before choosing a license for my code.

Now, I have a few issues as the R packages I used have all different types of licenses.

For example:

> packageDescription("ggplot2", fields="License")
[1] "MIT + file LICENSE"
> packageDescription("glmmTMB", fields="License")
[1] "AGPL-3"
> packageDescription("RColorBrewer", fields="License")
[1] "Apache License 2.0"

The list is quite long and I also have "GPL (>= 2.0)" and "GPL-3" . So, I am now very confused on how should I license my code, as I am aware that not in all cases the different licenses are compatible with each other. For example, GPL v2 is not always compliant with GPL v3 and vice-versa.

Is there a way I can understand this and avoid of making mistakes?

Edit

Sadly I also have used packages with GPL V2.

Answer 1

Let's look at the licenses you list:

• AGPL-3
• MIT
• Apache v2
• GPL v2+

MIT is a very permissive license and can be combined with either of the other three.

Apache v2 is a permissive license, but due to some patent indemnification clauses it is incompatible with GPL v2.

The AGPL v3 is like the GPLv3 but with the further requirement that its terms also apply when users interact with the software over a network interface.

Thus from the mentioned licenses, your only choice to distribute software with libraries using all mentioned licenses, is the AGPL v3.

Now that this is settled, you can use the list of licenses compatible with the (A)GPL v3 as provided by the FSF.

The most noteworthy license you can NOT combine your work with (and still legally distribute it), will be libraries which are under GPL v2 only (thus without the option to choose a newer version of the GPL). You will be able to include any library under any of the usual permissive licenses like MIT, BSD, etc.

May sound like a picky detail: you are not necessarily required to license your code under the AGPL v3 under these circumstances. But you may only distribute the whole software along with all the required libraries under AGPL v3 (as long as one of the required libraries is under AGPL v3).

This said: AGPL v3 is not a bad license for science, rather on the contrary: it requires attribution as any open source license, and it requires users to share their code modifications when they distribute it, and also share it when they offer a service with it, e.g. a website which interactively calculates stuff.

Further: if you do not distribute any complete software, you basically can choose any license - but the "wrong" choice might make it un-distributable for everyone else, too. As such my recommendation remains to always to use a license which allows distribution with all libraries used included. In your case that would be AGPL v3.