Typically, we rely on a number of external packages when programming in R. This document explores how those packages are installed and loaded.
When R is installed, it stores a number of paths (i.e. directories or folders) in its internal memory. These can be accessed with certain R functions.
For example, a key path is the “home” directory where the core R files live. You can find this path by running R.home():
R.home()## [1] "/Library/Frameworks/R.framework/Resources"You can then see which files and folders live here with the dir() function:
dir(R.home())## [1] "bin" "COPYING" "doc" "etc"
## [5] "fontconfig" "include" "Info.plist" "lib"
## [9] "library" "man1" "modules" "R"
## [13] "Rscript" "share" "SVN-REVISION" "tests"The folder structure may be familiar to Linux/Unix users. If not, that’s fine. We’re not concerned with the internal mechanics of R itself. Instead, we’ll focus on what happens with installed packages.
The command .libPaths() will print the folder(s) where R packages are downloaded (with install.packages()) or searched (with library()):
.libPaths()## [1] "/Users/gfiddyment/Library/R/3.3/library"
## [2] "/Library/Frameworks/R.framework/Versions/3.3/Resources/library"my.path = .libPaths()[1]Searching the folders in my.path will show one folder per installed package. We’ll look inside the folder for library(car):
dir(my.path)## [1] "abind" "acepack" "acs"
## [4] "akima" "alr4" "bdsmatrix"
## [7] "bindr" "bindrcpp" "bnlearn"
## [10] "car" "caret" "caTools"
## [13] "checkmate" "choroplethr" "coefplot"
## [16] "combinat" "corrplot" "coxme"
## [19] "ctv" "data.table" "DataCombine"
## [22] "devtools" "doParallel" "dplyr"
## [25] "dygraphs" "e1071" "effects"
## [28] "fivethirtyeight" "Formula" "gam"
## [31] "gbm" "geosphere" "GGally"
## [34] "ggmap" "ggplot2" "ggplot2movies"
## [37] "ggvis" "glmnet" "glue"
## [40] "gmp" "gridBase" "HH"
## [43] "Hmisc" "hms" "htmltab"
## [46] "htmlTable" "igraph" "igraphdata"
## [49] "irlba" "ISLR" "jpeg"
## [52] "klaR" "lattice" "latticeExtra"
## [55] "leaps" "lm.beta" "lmtest"
## [58] "mapproj" "maps" "mAr"
## [61] "MASS" "minqa" "mnormt"
## [64] "multcomp" "mvtnorm" "neuralnet"
## [67] "NMF" "packrat" "perturb"
## [70] "pkgconfig" "pkgmaker" "PKI"
## [73] "plogr" "pls" "plyr"
## [76] "png" "polspline" "prettyunits"
## [79] "pROC" "profvis" "progress"
## [82] "proto" "pryr" "psych"
## [85] "randomForest" "RCurl" "readr"
## [88] "registry" "reshape" "rgl"
## [91] "RgoogleMaps" "rJava" "rjson"
## [94] "RJSONIO" "rlang" "rmarkdown"
## [97] "Rmpfr" "rms" "rngtools"
## [100] "ROCR" "rpart" "rpart.plot"
## [103] "rsconnect" "rstan" "sand"
## [106] "sandwich" "sp" "StanHeaders"
## [109] "survey" "survival" "TH.data"
## [112] "tibble" "titanic" "tree"
## [115] "useful" "vcd" "VGAM"
## [118] "viridis" "WDI" "xlsx"
## [121] "xlsxjars" "XML"dir(paste(my.path, '/car', sep=''))## [1] "CITATION" "data" "DESCRIPTION" "doc" "help"
## [6] "html" "INDEX" "Meta" "NAMESPACE" "NEWS"
## [11] "R"We see there are sub-folders for data, documentation, and R functions:
dir(paste(my.path, '/car/data', sep=''))## [1] "Rdata.rdb" "Rdata.rds" "Rdata.rdx"dir(paste(my.path, '/car/doc', sep=''))## [1] "embedding.pdf" "embedding.R" "embedding.Rnw" "index.html"dir(paste(my.path, '/car/R', sep=''))## [1] "car" "car.rdb" "car.rdx"This is the typical setup, but there are also tools for making “environments” (i.e. folders with all code and dependent R packages together) that can easily be used across different computers. The main such tool is library(packrat). See http://rstudio.github.io/packrat/walkthrough.html for an introduction.