README.md

In Mullefa/locallib: Project isolation for R

locallib

virtualenv for R.

Overview

This package looks to replicate some of the functionality of virtualenv for R. Underneath the covers there's really just one thing going on: prepending a local library to the library trees within which packages are looked for. The user is trusted to determine whether they want to include non-base packages in their global library(s).

Getting started

Until locallib is available on CRAN, use devtools to install it:

if (!require("devtools")) {
  install.packages("devtools")
}

devtools::install_github("Mullefa/locallib")

To create your first project which utilises a local library, open R in a new project directory and load locallib:

library(locallib)

You will get a lot of warnings about packages being installed globally. If you want to remove these, so that all non-base packages are installed at the project level (recommended):

clean_global_libs()

Now you just have to create a local library and use it:

create_local_lib()
use_local_lib()

At this point, any installed packages will be installed into the local library (file path ./library). To get a snapshot of your local library:

freeze()

This will create a file specifying all the packages used in the project and their versions (file path ./pkgs.yaml). This is useful as now it is easy to replicate the local library:

thaw()

One last thing to mention is a pattern I use whenever starting a new project with its own local library. With a R console opened in the new project directory:

create_.Rprofile()

Now restart R. The created .Rprofile will activate the local library whenever the user is in interactive mode.

Use with multidplyr

The function cluster_use_local_lib() can be used to activate the locally activated local lib on each of the remote nodes e.g.

library(locallib)
library(dplyr)
library(multidplyr)

get_default_cluster %>% cluster_use_local_lib

Examples

Check out geohash-vis.

Mullefa/locallib documentation built on May 7, 2019, 5:12 p.m.