restore.library {groundhog}

Restore default library of packages, undoing all changes made by groundhog

Description

In a few seconds reverse changes made by groundhog to the default personal library (where R packages are usually installed into with installed.packages()). If you are just trying groundhog out for the first time, or you generally rely on base R's library() and want to use groundhog.library() sporadically then you may want to run restore.library() when you are done with your one-time use of groundhog. This will undo any and all changes made by groundhog to that library. In most circumstances, restoring a library takes a few seconds, even if the library has had 100s of package modifications.

Usage

restore.library(days = 0)

Arguments

days

an optional numeric argument used to choose among alternative restore points. When days is set, groundhog restores the personal library to the most recent restore point that is at least days days old. If dais is not set, groundhog restores to the most recent restore point overall. For example, if there are two restore points: one from today, and one from 7 days ago, running restore.library() would restore to the former, setting days=3 would restore to the latter, and setting days=8 would result in an error. days = -1 restores to the oldest restore point available.

Details

When groundhog installs a package, it installs it into groundhog's library (location of that library is obtainable with get.groundhog.folder()). Groundhog then immediately moves the installed package(s) (and their dependencies) to the default personal library (location of that library obtainable with: .libPaths()[1]). Altering the packages in the local folder is necessary for groundhog to work properly for two main reasons. First, R Studio often loads packages from that library before users run the code in a script, creating version conflicts that cannot be avoided when attempting to load other versions of those packages with groundhog. Second, R scripts often run processes in independent R sessions, for example when doing parallel processing. Those background processes will also look for packages in the default personal folder. Because the personal library can only hold one version of a given package, before moving new packages in, groundhog moves any existing other versions of those packages out, to another directory (a local archive). Those files are not deleted, just moved, making it easy and fast to recover. When the first change in the personal folder is made on a given calendar date, groundhog makes a list of all packages available in the personal folder before such change (saving a copy of the results from installed.packages(.libPaths()[1])), this saved file is referred to as a a 'restore point'. With restore.library() groundhog looks up a restore point, obtain the set of packages that used to be installed, and removes any packages installed by groundhog which are in the personal library but were not in that restore point; similarly, it moves back to the local library any packages removed by groundhog that were in the restore point but are not currently there. This process take a few seconds even for 100+ packages. Note that there is only one restore point per calendar date, so one effectively restores the personal library to how it was before any changes were made to it that day with groundhog. Restore points are saved permanently and can be restored at any point. The set of restore points available is stored in the hidden dataframe .available.restore.points. To choose among them use the days argument in restore.library. The default is to restore based on the most recent restore point, so if a user installs groundhog, tests it, and wants to undo all changes made by groundhog, the default behavior will achieve this goal. Note: restoring can take a few minutes if the groundhog folder is on a different drive from the default personal R library (e.g., two different hard drives) or if it is in Dropbox.

Examples

## Not run: 
restore.library()
restore.library(7)
restore.library(-1)

## End(Not run)