Why Drat? A Guest Post by Steven Pav

March 13, 2015
By

(This article was first published on Thinking inside the box , and kindly contributed to R-bloggers)

Editorial Note: The following post was kindly contributed by Steven Pav.

Why Drat?

After playing around with drat for a few days now, my impressions of it are best captured by Dirk’s quote:

It just works.

Demo

To get some idea of what I mean by this, suppose you are a happy consumer of R packages, but want access to, say, the latest, greatest releases of my distribution package, sadist. You can simply add the following to your .Rprofile file:

drat::add("shabbychef")

After this, you instantly have access to new releases in the github/shabbychef drat store via the package tools you already know and tolerate. You can use

install.package('sadists')

to install the sadists package from the drat store, for example. Similarly, if you issue

update.packages(ask=FALSE)

all the drat stores you have added will be checked for package updates, along with their dependencies which may well come from other repositories including CRAN.

Use cases

The most obvious use cases are:

  1. Micro releases. For package authors, this provides a means to get feedback from the early adopters, but also allows one to push small changes and bug fixes without burning through your CRAN karma (if you have any left). My personal drat store tends to be a few minor releases ahead of my CRAN releases.
  2. Local repositories. In my professional life, I write and maintain proprietary packages. Pushing package updates used to involve saving the package .tar.gz to a NAS, then calling something like R CMD INSTALL package_name_0.3.1.9001.tar.gz. This is not something I wanted to ask of my colleagues. With drat, they can instead add the following stanza to .Rprofile: drat:::addRepo('localRepo','file:///mnt/NAS/r/local/drat'), and then rely on update.packages to do the rest.

I suspect that in the future, drat might be (ab)used in the following ways:

  1. Rolling your own vanilla CRAN mirror, though I suspect there are better existing ways to accomplish this.
  2. Patching CRAN. Suppose you found a bug in a package on CRAN (inconceivable!). As it stands now, you email the maintainer, and wait for a fix. Maybe the patch is trivial, but suppose it is never delivered. Now, you can simply make the patch yourself, pick a higher revision number, and stash it in your drat store. The only downside is that eventually the package maintainer might bump their revision number without pushing a fix, and you are stuck in an arms race of version numbers.
  3. Forgoing CRAN altogether. While some package maintainers might find this attractive, I think I would prefer a single huge repository, warts and all, to a landscape of a million microrepos. Perhaps some enterprising group will set up a CRAN-like drat store on github, and accept packages by pull request (whether github CDN can or will support the traffic that CRAN does is another matter), but this seems a bit too futuristic for me now.

My wish list

In exchange for writing this blog post, I get to lobby Dirk for some features in drat:

  1. I shudder at the thought of hundreds of tiny drat stores. Perhaps there should be a way to aggregate addRepo commands in some way. This would allow curators to publish their suggested lists of repos.
  2. Drat stores are served in the gh-pages branch of a github repo. I wish there were some way to keep the index.html file in that directory reflect the packages present in the sources. Maybe this could be achieved with some canonical RMarkdown code that most people use.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

 

To leave a comment for the author, please follow the link and comment on their blog: Thinking inside the box .

R-bloggers.com offers daily e-mail updates about R news and tutorials on topics such as: Data science, Big Data, R jobs, visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...



If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.

Search R-bloggers


Sponsors

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)