deref: dereferencing references

In ref: References for R

Description

This functions allow to access a referenced object. deref(ref) returns the object, and deref(ref) <- value assigns to the referenced object.

Usage

deref(ref)
deref(ref) <- value
#the following does not pass R CMD CHECK
#deref<-(ref, value)
#deref(ref)[1] <- value  # subsetted assignment appears to be inefficent in S+.

Arguments

ref	a reference as returned by ref or as.ref
value	a value to be assigned to the reference

Details

deref and deref<- provide convenient access to objects in other environments/frames. In fact they are wrappers to get and assign. However, convenient does not neccessarily means efficient. If performance is an issue, the direct use of new.env, substitute and eval may give better results. See the examples below.

Value

deref returns the referenced object.
"deref<-" returns a reference to the modified object, see ref.

Note

Subsetted assignment appears to be inefficent in S+. Note the use of substitute in the examples.

Author(s)

Jens Oehlschl<e4>gel

References

Writing R Extensions

See Also

ref, as.ref, get, assign, substitute, eval

Examples

  # Simple usage example
  x <- cbind(1:5, 1:5)          # take some object
  rx <- as.ref(x)               # wrap it into a reference
  deref(rx)                     # read it through the reference
  deref(rx) <- rbind(1:5, 1:5)  # replace the object in the reference by another one
  deref(rx)[1, ]                # read part of the object
  deref(rx)[1, ] <- 5:1         # replace part of the object
  deref(rx)                     # see the change
  cat("For performance test examples see ?deref\n")

 ## Not run: 
  ## Performance test examples showing actually passing by reference
  # define test matrix size of nmatrix by nmatrix
  nmatrix <- 1000
  # you might want to use less loops in S+
  # you might want more in R versions before 1.8
  nloop   <- 10     

  # Performance test using ref
  t1 <- function(){ # outer function
    m <- matrix(nrow=nmatrix, ncol=nmatrix)
    a <- as.ref(m)
      t2(a)
    m[1,1]
  }
  # subsetting deref is slower (by factor 75 slower since R 1.8 compared to previous versions
  # , and much, much slower in S+) ...
  t2 <- function(ref){
    cat("timing", timing.wrapper(
      for(i in 1:nloop)
        deref(ref)[1,1] <- i
    ), "\n")
  }
  if (is.R())gc()
  t1()
  # ... than using substitute
  t2 <- function(ref){
    obj <- as.name(ref$name)
    loc <- ref$loc
    cat("timing", timing.wrapper(
      for(i in 1:nloop)
        eval(substitute(x[1,1] <- i, list(x=obj, i=i)), loc)
    ), "\n")
  }
  if (is.R())gc()
  t1()


  # Performance test using Object (R only)
  # see Henrik Bengtsson package(oo)
  Object <- function(){
    this <- list(env.=new.env());
    class(this) <- "Object";
    this;
  }
  "$.Object" <- function(this, name){
    get(name, envir=unclass(this)$env.);
  }
  "$<-.Object" <- function(this, name, value){
    assign(name, value, envir=unclass(this)$env.);
    this;
  }
  # outer function
  t1 <- function(){
    o <- Object()
    o$m <- matrix(nrow=nmatrix, ncol=nmatrix)
      t2(o)
    o$m[1,1]
  }
  # subsetting o$m is slower ...
  t2 <- function(o){
    cat("timing", timing.wrapper(
      for(i in 1:nloop)
        o$m[1,1] <- i
    ), "\n")
  }
  if (is.R())gc()
  t1()
  # ... than using substitute
  t2 <- function(o){
    env <- unclass(o)$env.
    cat("timing", timing.wrapper(
      for(i in 1:nloop)
        eval(substitute(m[1,1] <- i, list(i=i)), env)
    ), "\n")
  }
  if (is.R())gc()
  t1()

  
## End(Not run)

Example output

dim(refdata) and dimnames(refdata) no longer allow parameter ref=TRUE, use dim(derefdata(refdata)), dimnames(derefdata(refdata)) instead
     [,1] [,2]
[1,]    1    1
[2,]    2    2
[3,]    3    3
[4,]    4    4
[5,]    5    5
[1] 1 2 3 4 5
     [,1] [,2] [,3] [,4] [,5]
[1,]    5    4    3    2    1
[2,]    1    2    3    4    5
For performance test examples see ?deref
         used (Mb) gc trigger (Mb) max used (Mb)
Ncells 257577 13.8     460000 24.6   350000 18.7
Vcells 456287  3.5    1023718  7.9   786431  6.0
timing 0.068 
[1] 10
         used (Mb) gc trigger (Mb) max used (Mb)
Ncells 314691 16.9     592000 31.7   350000 18.7
Vcells 584356  4.5    2216018 17.0  2584528 19.8
timing 0.005 
[1] 10
         used (Mb) gc trigger (Mb) max used (Mb)
Ncells 315554 16.9     592000 31.7   371417 19.9
Vcells 586264  4.5    2216018 17.0  2584528 19.8
timing 0.044 
[1] 10
         used (Mb) gc trigger (Mb) max used (Mb)
Ncells 316024 16.9     592000 31.7   371417 19.9
Vcells 587355  4.5    2191376 16.8  2587558 19.8
timing 0.002 
[1] 10

ref documentation built on May 30, 2017, 2:19 a.m.