Last data update: 2014.03.03

R: Byte Code Compiler
compileR Documentation

Byte Code Compiler

Description

These functions provide an interface to a byte code compiler for R.

Usage

cmpfun(f, options = NULL)
compile(e, env = .GlobalEnv, options = NULL)
cmpfile(infile, outfile, ascii = FALSE, env = .GlobalEnv,
        verbose = FALSE, options = NULL)
loadcmp(file, envir = .GlobalEnv, chdir = FALSE)
disassemble(code)
enableJIT(level)
compilePKGS(enable)
getCompilerOption(name, options)
setCompilerOptions(...)

Arguments

f

a closure.

options

list of named compiler options

env

the top level environment for the compiling.

file,infile,outfile

pathnames; outfile defaults to infile with a .Rc extension in place of any existing extension.

ascii

logical; should the compiled file be saved in ascii format?

verbose

logical; should the compiler show what is being compiled

envir

environment to evaluate loaded expressions in.

chdir

logical; change directory before evaluation?

code

byte code expression or compiled closure

e

expression to compile

level

integer; the JIT level to use

enable

logical; enable compiling packages if TRUE

name

character string; name of option to return

...

named compiler options to set

Details

The function cmpfun compiles the body of a closure and returns a new closure with the same formals and the body replaced by the compiled body expression.

compile compiles an expression into a byte code object; the object can then be evaluated with eval.

cmpfile parses the expression in infile, compiles them, and writes the compiled expressions to outfile. If outfile is not provided, it is formed from infile by replacing or appending a .Rc suffix.

loadcmp is used to load compiled files. It is similar to sys.source, except that its default loading environment is the global environment rather than the base environment.

disassemble produces a printed representation of the code that may be useful to give a hint of what is going on.

enableJIT enables or disables just-in-time (JIT) compilation. JIT is disabled if the argument is 0. If enable is 1 then closures are compiled before their first use. If enable is 2, then in addition closures are also compiled before they are duplicated (useful for some packages, like lattice, that store closures in lists). If enable is 3 then in addition all loops are compiled before they are executed. JIT level 3 requires optimization level 2 or 3. JIT can also be enabled by starting R with the environment variable R_ENABLE_JIT set to one of these values. Calling enableJIT with a negative argument returns the current JIT level.

compilePKGS enables or disables compiling packages when they are installed. This requires that the package use lazy loading as compilation occurs as functions are written to the lazy loading data base. This can also be enabled by starting R with the environment variable R_COMPILE_PKGS set to a positive integer value.

Currently the compiler warns about a variety of things. It does this by using cat to print messages. Eventually this should use the condition handling mechanism.

The options argument can be used to control compiler operation. There are currently three options: optimize, suppressAll, and suppressUndefined. optimize specifies the optimization level, which can be an integer form 0 to 3. suppressAll should be a scalar logical; if TRUE no messages will be shown. suppressUndefined can be TRUE to suppress all messages about undefined variables, or it can be a character vector of the names of variables for which messages should not be shown.

getCompilerOption returns the value of the specified option. The default value is returned unless a value is supplied in the options argument; the options argument is primarily for internal use. setCompilerOption sets the default option values. It returns a named list of the previous values.

Calling the compiler a byte code compiler is actually a bit of a misnomer: the external representation of code objects currently uses int operands, and when compiled with gcc the internal representation is actually threaded code rather than byte code.

Author(s)

Luke Tierney

Examples

# a simple example
f <- function(x) x+1
fc <- cmpfun(f)
fc(2)
disassemble(fc)

# old R version of lapply
la1 <- function(X, FUN, ...) {
    FUN <- match.fun(FUN)
    if (!is.list(X))
	X <- as.list(X)
    rval <- vector("list", length(X))
    for(i in seq(along = X))
	rval[i] <- list(FUN(X[[i]], ...))
    names(rval) <- names(X)		  # keep `names' !
    return(rval)
}
# a small variation
la2 <- function(X, FUN, ...) {
    FUN <- match.fun(FUN)
    if (!is.list(X))
	X <- as.list(X)
    rval <- vector("list", length(X))
    for(i in seq(along = X)) {
        v <- FUN(X[[i]], ...)
        if (is.null(v)) rval[i] <- list(v)
        else rval[[i]] <- v
    }
    names(rval) <- names(X)		  # keep `names' !
    return(rval)
}
# Compiled versions
la1c <- cmpfun(la1)
la2c <- cmpfun(la2)
# some timings
x <- 1:10
y <- 1:100

system.time(for (i in 1:10000) lapply(x, is.null))
system.time(for (i in 1:10000) la1(x, is.null))
system.time(for (i in 1:10000) la1c(x, is.null))
system.time(for (i in 1:10000) la2(x, is.null))
system.time(for (i in 1:10000) la2c(x, is.null))
system.time(for (i in 1:1000) lapply(y, is.null))
system.time(for (i in 1:1000) la1(y, is.null))
system.time(for (i in 1:1000) la1c(y, is.null))
system.time(for (i in 1:1000) la2(y, is.null))
system.time(for (i in 1:1000) la2c(y, is.null))

Results


R version 3.3.1 (2016-06-21) -- "Bug in Your Hair"
Copyright (C) 2016 The R Foundation for Statistical Computing
Platform: x86_64-pc-linux-gnu (64-bit)

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.

R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.

> library(compiler)
> png(filename="/home/ddbj/snapshot/RGM3/R_rel/result/compiler/compile.Rd_%03d_medium.png", width=480, height=480)
> ### Name: compile
> ### Title: Byte Code Compiler
> ### Aliases: compile cmpfun cmpfile loadcmp disassemble enableJIT
> ###   compilePKGS getCompilerOption setCompilerOptions
> ### Keywords: programming
> 
> ### ** Examples
> 
> # a simple example
> f <- function(x) x+1
> fc <- cmpfun(f)
> fc(2)
[1] 3
> disassemble(fc)
list(.Code, list(8L, GETVAR.OP, 1L, LDCONST.OP, 2L, ADD.OP, 0L, 
    RETURN.OP), list(x + 1, x, 1))
> 
> # old R version of lapply
> la1 <- function(X, FUN, ...) {
+     FUN <- match.fun(FUN)
+     if (!is.list(X))
+ 	X <- as.list(X)
+     rval <- vector("list", length(X))
+     for(i in seq(along = X))
+ 	rval[i] <- list(FUN(X[[i]], ...))
+     names(rval) <- names(X)		  # keep `names' !
+     return(rval)
+ }
> # a small variation
> la2 <- function(X, FUN, ...) {
+     FUN <- match.fun(FUN)
+     if (!is.list(X))
+ 	X <- as.list(X)
+     rval <- vector("list", length(X))
+     for(i in seq(along = X)) {
+         v <- FUN(X[[i]], ...)
+         if (is.null(v)) rval[i] <- list(v)
+         else rval[[i]] <- v
+     }
+     names(rval) <- names(X)		  # keep `names' !
+     return(rval)
+ }
> # Compiled versions
> la1c <- cmpfun(la1)
> la2c <- cmpfun(la2)
> # some timings
> x <- 1:10
> y <- 1:100
> ## No test: 
> system.time(for (i in 1:10000) lapply(x, is.null))
   user  system elapsed 
  0.044   0.000   0.044 
> system.time(for (i in 1:10000) la1(x, is.null))
   user  system elapsed 
  0.296   0.004   0.304 
> system.time(for (i in 1:10000) la1c(x, is.null))
   user  system elapsed 
    0.2     0.0     0.2 
> system.time(for (i in 1:10000) la2(x, is.null))
   user  system elapsed 
  0.260   0.000   0.257 
> system.time(for (i in 1:10000) la2c(x, is.null))
   user  system elapsed 
  0.108   0.000   0.107 
> system.time(for (i in 1:1000) lapply(y, is.null))
   user  system elapsed 
  0.028   0.000   0.030 
> system.time(for (i in 1:1000) la1(y, is.null))
   user  system elapsed 
  0.192   0.000   0.192 
> system.time(for (i in 1:1000) la1c(y, is.null))
   user  system elapsed 
  0.068   0.000   0.067 
> system.time(for (i in 1:1000) la2(y, is.null))
   user  system elapsed 
   0.12    0.00    0.12 
> system.time(for (i in 1:1000) la2c(y, is.null))
   user  system elapsed 
  0.040   0.000   0.043 
> ## End(No test)
> ## Don't show: 
> for (i in 1:10000) lapply(x, is.null)
> for (i in 1:10000) la1(x, is.null)
> for (i in 1:10000) la1c(x, is.null)
> for (i in 1:10000) la2(x, is.null)
> for (i in 1:10000) la2c(x, is.null)
> for (i in 1:1000) lapply(y, is.null)
> for (i in 1:1000) la1(y, is.null)
> for (i in 1:1000) la1c(y, is.null)
> for (i in 1:1000) la2(y, is.null)
> for (i in 1:1000) la2c(y, is.null)
> ## End(Don't show)
> 
> 
> 
> 
> dev.off()
null device 
          1 
>