Separating main statistical content from coding

[poldrack]

This issue is meant to describe a major planned change for the organization of the book.

Motivation:
The proposed change is driven by two primary issues:

1. Later this year I am planning to write Python-based companion chapters to serve in parallel with the R-based companion chapters. Ultimately I would like the main statistical text to be fully language-agnostic so that the reader could choose between the R or Python companions.
2. I am in discussion with a publisher to publish a low-cost paperback edition of the main statistical text. The main text will remain available online under the same license as present (CC-BY-NC), but I would like to separately release the statistical companions under a less restrictive license (such as CC-BY).

Proposal
I propose to split the current book out into two separate books - one for the main statistical material, and a companion for the R chapters. This way they can be built separately, so that the more frequent updates to the companion will not require building the full main text.

I think the best way to do this would be to separate the R chapters into a separate repo, with an additional separate repo for the python chapters to be generated later.

Comments and suggestions welcome.

[complexbrains]

Fantastic news! I am happy to help with the Python conversion and organization into two separate books.

I was also thinking it might be nice to provide the code of the estimations/plots embedded within the text material separately, maybe as separate scripts or maybe again within those following practice chapters.

Since going through them also teach a lot and having them separately might be very helpful for those are not familiar with the GitHub practices or who do not want to hassle with picking them through the context. They are there already, so I believe it would not be too much work for us. Just a thought.

[poldrack]

that's a great idea to flesh out the actual code that is used in the chapter in more detail in the companions!

[complexbrains]

If there are no license issues atm, we can start working on the companion repo and continue maintaining the current repo as it is. Once you decide to separate them completely, it will be just up to us to organize the current repo and make the regarding changes in the automatic production of the online version of the book via CircleCi if there is.

I guess since you still teach, I believe it will be at the end of the term sometime, hopefully by then at least we can have the companion repo is as ready as possible. Happy to help with that!

[poldrack]

I'm still teaching for two more weeks so I don't want to make any changes until after that...

[mgxd]

Sounds like a good idea, one suggestion would be to have the main repository (this one?) consist of only the statistical material and then include the R/Python companions as git submodules.