Go + R + Go + Emacs = Rbooks

[Jason E. Aten]

https://github.com/glycerine/rbook

I've open sourced my rbook project. I built rbook for myself to save data

analysis in R to a file which serves as a lab notebook. I call the 

program rbook, and the saved files rbooks.  

I regularly use emacs and ESS (Emacs-Speaks-

Statistics) for R based data analysis, and I wanted a way

to record the sequence of plots and R output to a web browser.

With rbooks, I can easily revisit and scroll through the analysis.

All R commands and output or recorded, plots are shown 

in-line, and comments can be inserted.  R can run out of 

memory and crash, so it is nice to be able to have a persistent

session log--with graphics--to recovery and start again from.

Using rbook is similar to Jupyter Notebooks and Jupyter Hub, but

this is a completely independent project. I prefer emacs and ESS to Jupyter,

as it is much faster to work in. I have ctrl-n bound to step through

code, making debugging R code seamless.  From my .emacs:

(global-set-key "\C-n" 'ess-eval-line-and-step)

The rbook project also demonstrates that it is viable to have a Go program

host DLLs that are also written in Go. By controlling the signal

handling, the two Go runtimes can coexist.

Rbook is a Go program uses my R embedding library ( https://github.com/glycerine/embedr ) to load R as a DLL. 

In turn, R can load other Go code that has been built to be an R package DLL.

The embedr and rbook Go code does the signal handling settings

needed to make this possible.  This took a long time to figure out,

but it has been stable and working for a year now.

I find it incredibly useful to be able to call into my Go code from R.

Enjoy.