3 February 2013

InstallOldPackages: a repmis command for installing old R package versions

A big problem in reproducible research is that software changes. The code you used to do a piece of research may depend on a specific version of software that has since been changed. This is an annoying problem in R because install.packages only installs the most recent version of a package. It can be tedious to collect the old versions.

On Toby Dylan Hocking's suggestion, I added tools to the repmis package so that you can install, load, and cite specific R package versions. It should work for any package version that is stored on the CRAN archive (http://cran.r-project.org).

To only install old package versions use the new repmis command InstallOldPackages. For example:

# Install old versions of the e1071 and gtools packages.

# Create vectors of the package names and versions to install
# Note the names and version numbers must be in the same order
Names <- c("e1071", "gtools")
Vers <- c("1.6", "2.6.1")

# Install old package versions into the default library
InstallOldPackages(pkgs = Names, versions = Vers)

You can also now have LoadandCite install specific package versions:

# Install, load, and cite specific package versions

# Create vectors of the package names and versions to install
# Note the names and version numbers must be in the same order
Names <- c("e1071", "gtools") 
Vers <- c("1.6", "2.6.1")

# Run LoadandCite
LoadandCite(pkgs = Names, versions = Vers, install = TRUE, file = "PackageCites.bib")

See this post for more details on LoadandCite.


I intend to continue improving these capabilities. So please post any suggestions for improvement (or report any bugs) at on the GitHub issues page.


M Edward Borasky said...

This is kind of interesting to me, and at the same time there's a bit of a philosophical issue I have.

Some background: I collect and distribute tools for computational journalism. See http://znmeb.github.com/CompJournoStick. The core of these tool collections is a Linux desktop, R and in some cases RStudio.

Recently, two of the packages that have been in past versions of this tool set, BARD and RcmdrPlugin.TextMining, went out of the main CRAN repository and are only available from the archive repository.

As a distributor, I need a way to detect when this happens without running my install script. And I need to know whether my users actually want a package that's no longer being maintained or has been kicked out of the main CRAN repository for some reason.

So I will definitely check this tool out and see if it can help me, but at the same time I'm not sure I'd want to facilitate use of "obsolete" or "unmaintained" software.

pirategrunt.com said...

This may be a silly question, but can repmis also sort out re-installation of packages when I upgrade my version of R? I recently upgraded and found that many packages had disappeared. I can see them out there under a folder with the old R version number, but I've had to reinstall in order to make them available.

Again, there's probably some basic bit of idiocy that I committed during the install. Just curious if there's an easy way to undo it.

Christopher Gandrud said...

Re M Edward Borasky:

I agree that you should generally be using the most updated version of a package/maintained packages for your research.

I intend InstallOldPackages and the similar functionality in LoadandCite to be used for replication purposes only.

When a piece of research is under active development researchers should use LoadandCite without specifying the package version. If install = TRUE then only the most recent versions of the packages will be installed from CRAN.

When a researcher releases a final replication version of their Sweave or knitr file then they should specify the package versions in LoadandCite. This help make the code in their file run as intended during replication.

Christopher Gandrud said...

Re pirategrunt.com

This has to do with where your library path is. By default each new major version resets the path.

You can change the library path: see this Stack Exchange page for more details: http://stackoverflow.com/questions/2615128/where-does-r-store-packages.

Toby Dylan Hocking said...

Great work. Your example installation of old packages gtools and e1071 worked for me.

But do you have any ideas about how to resolve the chicken-and-egg problem? i.e. what version of repmis should be required, and how to indicate that?

Christopher Gandrud said...

Ha, yeah there is definitely a chicken and egg problem.

I'll have to think about what can be done. But at the very least LoadandCite consolidates the issue of having to manually update replication code into one command, rather than having to go through a whole analysis and update all of the packages and/or syntax that may have changed.

Anonymous said...

Talking about replicating: the change in the theming system after ggplot2 0.8.9 broke a *lot* of code for me.

When I wanted to replicate some graphs, I was so annoyed that I returned to the old version by hand.

Thanks for drawing my attention to {repmis} via R-bloggers!