I am sure this has been brought up before but I couldn't find the precise answer... So forgive me if this is pinned somewhere.
Tidyverse is MIT licensed but relies on GPL packages (like Rcpp). This means I could not use tidyverse (specifically, DPLYR) for a closed-source project that is accessible by API, correct?
This seems to limit R / Tidyverse's practicality in enterprise. Note that I do LOVE Rstudio / Shiny and especially dplyr. However, it is hard to convince a corporate environment to invest in R while Python is generously licensed. What am I missing?
It takes a bit of direct research, and don't trust summaries from non-specialists such as myself. I am not a lawyer, not giving legal advice, and not affiliated with RStudio. I am not claiming this reply is an answer. I am just trying to point out the topics I think one would need to research to answer your question. Or at least things to look for in answers.
To orient the bit. I would suggest researching all of the systems and licenses separately. In particular R, RStudio, Shiny, and tidyverse each may or may not bring in different license issues.
R itself is licensed GPL2/3. So if one doesn't want GPL software then it doesn't matter what R packages you use as R already brings in GPL2/3 rights and responsibilities.
The GPL2/3 license has conditions about what a GPL2/3 item can be distributed with. One of these stats that things that are distributed with, link to, or even built on APIs must themselves have a compatible license when distributed. The MIT license is considered compatible with GPL. So the tidyverse likely isn't introducing many more obligations than just using R, as the MIT license is considered fairly permissive. However it also isn't freeing parties from the GPL2/3 conditions.
The open-source distributions of RStudio and Shiny are under an AGPL v3 license. This license is significant in that it grants rights to downstream users even without code distribution. Roughly: using the code grants the same rights to the user as distributing the code would, including a right to the source code under an AGPL v3 compatible license. This license is very different than the GPL2/3 or MIT licenses, so requires its own study.
Thank you for the thoughtful response. I know this helped clarify a few items for my team. I wonder if this opens up a broader question about R's future in the private sector. Shiny and RStudio have a great way to solve this (pay for a commercial license) - but the broader ecosystem concerns me. Obviously, not to be solved here.
GPL 2 is reasonably friendly. For example, Google allows engineers to use GPL 2 software in their internal systems.
GPL 3 and AGPL is fairly unfriendly for companies. Google does NOT allow use of these licenses in any systems that touch or use data from their core engineering code. That is because these licenses are considered “viral” by many lawyers. Using them in hosted services could legally embroil them in having to release their internal code for other systems that are used with this code.
I know this sounds strange, but it is the best of my understanding.
MIT License is very broad and acceptable in almost any application.