[R] Best way to study internals of R ( mix of C, C++, Fortran, and R itself)?

[Duncan Murdoch]

On 21/11/2017 2:14 PM, Robert Wilkins wrote:
> How difficult is it to get a good feel for the internals of R, if you want
> to learn the general code base, but also the CPU intensive stuff ( much of
> it in C or Fortran?) and the ways in which the general code and the CPU
> intensive stuff is connected together?

That's a pretty difficult question to answer.  How hard compared to what?

> 
> R has a very large audience, but my understanding is that only a small
> group have a good understanding of the internals (and some of those will
> eventually move on to something else in their career, or retire
> altogether).

That's true, but the good news is that there are people who know the 
internals now who didn't know them 5 or 10 years ago.  So there is 
renewal happening.  And there are a number of independent 
implementations of the language or subsets of it; see the Wikipedia 
article <https://en.wikipedia.org/wiki/R_(programming_language)>.

> While I'm at it, a second question: 15 years ago, nobody would ever offer a
> job based on R skills ( SAS, yes, SPSS, maybe, but R skills, year after
> year, did not imply job offers). How much has that changed, both for R and
> for NumPy/Pandas/SciPy ?
> 

The web page <http://r4stats.com/articles/popularity/> is fairly up to 
date.  It doesn't say what things were like in 2002, but in early 2017, 
the ranking was Python > R > SAS in the count of job ads in data 
science.  In 2012 it was SAS > Python > R (but R and Python were very 
close).

Duncan Murdoch