Does the documentation or language definition for R remark on the intended usage of R's integers?

Question

A common comment that I see made about R's integer type is that it's only really intended for communication with C code. Do any statement like this appear in any official part of R's documentation? I often catch myself making vectors like integer(10) under the impression that they'll be more efficient for my purposes, only to remember this folklore and reconsider if I should ever be using integers for code that never tries to communicate with C code.

Genevivegenevra

R docs are here, although it's easier to search everything at once if you get the whole source tree (by downloading the source tarball or checking out a copy from the SVN repository) – Vegetarian

@Henrik Ah, you're talking about "Integer vectors exist so that data can be passed to C or Fortran code which expects them, and so that (small) integer data can be represented exactly and compactly." in ?integer? Yes, I was a fool to miss that. – Genevivegenevra

Related: Benefits of using integer values for constants rather than numeric values (e.g. 1L vs 1) in R – Gastroenterology

Answer 1

I don't think so. This folklore probably comes from the fact that R is pretty loose about typing and coercion, so it's easy to end up with a floating-point variable by accident.

Integer types certainly save memory:

> object.size(seq(1e8))
400000048 bytes
> object.size(seq(1e8)+0.1)
800000048 bytes

I haven't tried benchmarking to see if R uses faster routines for integer vs floating-point arithmetic, but you could.

I haven't looked carefully through all of R's documentation, but the only slightly relevant comment that turns up in a full-text search for "integer" in the R language definition is:

In most cases,the difference between an integer and a numeric value will be unimportant as R will do the right thing when using the numbers. There are, however, times when we would like to explicitly create an integer value for a constant. We can do this by calling the function as.integer or using various other techniques ...

I did a grep integer *.texi in the doc/manual directory of the R source tree and didn't (in a quick skim) notice anything else that looked relevant.

Vegetarian

Answer 2

Following Ben Bolker's advice, I checked the seven R manuals. In addition to Ben's answer, I found the following:

For most purposes the user will not be concerned if the “numbers” in a numeric vector are integers, reals or even complex. Internally calculations are done as double precision real numbers, or double precision complex numbers if the input data are complex.
_{An Introduction to R Section 2.2}

Writing R Extensions gives lots of guidance for making R communicate with C and Fortran, but it doesn't say anything about the integer typing's intent.

The last place to check is the Full Reference Manual. You would have to be mad to do so - the word "integer" occurs over 1000 times. However, a quick look at the index reveals the documentation for the integer class. This gives us the answer is such plain English that I should not be forgiven for having missed it:

Integer vectors exist so that data can be passed to C or Fortran code which expects them, and so that (small) integer data can be represented exactly and compactly.

Genevivegenevra