Skip to content

Latest commit

 

History

History
79 lines (61 loc) · 2.54 KB

REPL.md

File metadata and controls

79 lines (61 loc) · 2.54 KB
Raw

REPL - Read, Evaluate, Print, Loop.

A REPL is a loop (the L part) that

  • Reads input and parses it into an R command(s)
  • Evaluates the command(s)
  • Prints the result and starts over again.

Why do we care? Partially because it helps to understand the nature of an error which might be a syntax error (at parse time) or run/evaluation-time semantic errors (e.g. no such variable, wrong input to a function).

We start an interpreter (R, python, MATLAB, etc.) and it gives us a prompt, waiting for our command. It is awaiting and reading our input. When we enter a character, it processes it potentially doing something such as suggesting auto-completion. The R part in the REPL comes in when we hit the return key. The REPL attempts to parse the text. There are three possibilities:

  1. the line makes complete sense as a command,
  2. contains a syntax error, or
  3. forms an incomplete command that makes partial sense and that needs to be completed.

There are two steps in an R command - checking it is syntactically correct, and then if it makes sense - evaluating it.

R parses the text of the command to see if it is syntactically correct. If it is not, we get a syntax error. There are so many ways to write a command that is not syntactically correct. For example,

lapply(x mean)
2pi = 1
x[[2][1]
lapply(x mean na.rm = TRUE)

If the command is syntatically complete and legitimate, i.e., makes sense syntactically, R tries to evaluate the expression. The rules for evaluation are somewhat involved, but basically we find variables along the search path, and we make function calls. That's pretty much all there is to the R language.

It may seem that there is a lot more to the language, e.g. math operators, logic operators, sequence operator (a:b), subsetting ([, [[, $), assignments, for and while loops, if statements. But in fact, these are all function calls. That is one of the elegant aspects of the R language. Master function calls and you master the language. However, function calls are rich.

The parse() and eval() functions

The REPL steps happen somewhat transparently. R reads the input and evaluates the command and prints. The REPL doesn't mention parsing. Sometimes it can be useful to actually parse the command explicitly. We can use

e = parse("nameOfFile")

or

e = parse(text = "text of R command")

This can help us verify the code is syntactically correct.

We can evaluate the parsed expressions with

eval(e)

And we can control the environment in which it is evaluated.