Quick comparison of
	"bootstrap"   Efron & Tibshirani
	"boot"        Canty, Davison & Hinkley
	"S+Resample"  S+ only
	"resample"    R only

--------------------------------------------------
The `bootstrap' library contains code used in Efron & Tibshirani,
"An Introduction to the Bootstrap", 1993.
* limited scope, with 8 functions intended to be called by users, 11 total
* no print, plot, or summary methods methods
* user must write functions to calculate their statistic in different forms
  expected by the eight functions,
* short and simple code, good for someone who wants to write their own
  code using this as an example.

# Example:
b <- bootstrap(fuel.frame$Fuel, 1000, mean)
b  # raw results
c(original = mean(fuel.frame$Fuel),
  mean = mean(b$thetastar),
  bias = mean(b$thetastar) - mean(fuel.frame$Fuel),
  stderr = stdev(b$thetastar))
hist(b$thetastar)
qqnorm(b$thetastar)
my.weighted.mean <- function(p, x) sum(x*p)/sum(p)
abcnon(fuel.frame$Fuel, my.weighted.mean)$limits

--------------------------------------------------
The `boot' library contains code by Angelo Canty for Davison & Hinkley,
"Bootstrap Methods and their Applications", 1997.
* More extensive scope, 75 functions total
* has 4 print methods and 2 plot methods
* user must write functions to calculate their statistic in different forms
* longer code, handles more cases (e.g. stratified sampling), still
  suitable for people to use as an example to modify

# Example:
my.mean <- function(x,i) mean(x[i])
b <- boot(fuel.frame$Fuel, my.mean, 1000, stype="i")
b  # formatted summary of results
plot(b)
my.weighted.mean <- function(x, w) sum(w*x)/sum(w)
abc.ci(fuel.frame$Fuel, my.weighted.mean)


--------------------------------------------------
The S+Resample library has been available since June 2002.
* Wide scope, 223 functions
  bootstrap, jackknife, permutation tests, cross validation,
  bootstrapping prediction errors, parametric and smoothed bootstrapping,
  tilting,
  bootstrap and permutation test for difference of statistic on 2 samples,
* variety of resampling methods: stratified, by subject, finite population,
  block bootstrap, subsampling,
* has print, plot, qqnorm, and summary methods,
* easier to use interface, can give function or expression, and use
  variables in a data frame,
* ugly code - you don't want to modify this for your own use,
* many routines coded in C for speed,
* on Windows, has more extended graphical interface.

# Example:
b <- bootstrap(fuel.frame$Fuel, mean)
b <- bootstrap(fuel.frame, mean(Fuel))  # alternate form
b  # formatted summary of results
plot(b)
qqnorm(b)
limits.bca(b)

# The library also provides support for two-sample bootstrapping
# and permutation tests, e.g.
bootstrap2(fuel.frame, mean(Weight), treatment = Type == "Sporty")

--------------------------------------------------
The `resample' package for R has been available since 2014, on CRAN.

This is a limited copy of S+Resample, focusing on ease of use for the
most common tasks -- one- and two-sample bootstrap and permutation tests.
I'll add to this over time.

# Example: (same as S+Resample)