Renv
Like conda in python, R has a package management tool (albeit less advanced), called renv. I personally recommend having these for any package / set of tools you develop. I’m going to introduce it relative to a shiny app on our shiny server<ref>In the future I can imagine someone extending to this commentary with commentary relative to a R package or at least a github repository - not just a set of files.</ref>.
1. Motivation
renv for R and conda for python are useful tools (Python virtual environments). They allow the user to more easily share code and not work about dependency issues for different people (recommended for consultant projects and group projects). Additionally, here at CMU Statistics and Data Science, if you’ve used the servers, you’ve probably experienced the server crash and remove all your R libraries (or you’ve updated Rstudio and a similar thing has occurred). Enter the renv package. This package stores all your libraries, their versions and their dependencies.
2. How to think about renv workflow
Below you’ll see a getting started example walkthrough. Something that may catch your eye is that the package updating and saving has similar structures and function names to git workflow (which you learned in Stat Computing).
3. Getting started
3.1 Example directory
Let’s suppose you already have a set of files going, (for this example, a shiny app). Specifically, in my Shiny app repository 36-350_shiny_files I have:
. ├── 36350 │ ├── 350_syllabus.pdf │ ├── Homework │ │ ├── hw_1-assign.Rmd │ │ └── ... │ ├── Labs │ │ ├── lab_1-assign.Rmd │ │ └── ... │ └── Lectures │ ├── 1.1_intro.Rmd │ └── ... └── app.R
3.2 .renvignore (Ignoring files)
Basically this is just a shiny-based file sharing app (don’t ask why it exists). It contains Rmd files and an app.R file, but I really only want to make sure that I keep all my libraries need to make app.R run correctly. To hide these other files, I create a .renvignore file and make it look like the following.
36350/*
The syntax for this file is the exact same as a .gitignore file and if you don’t make the .renvignore file it will just use a .gitignore file if you have one (see Documenation of Ignoring files for more details).
3.3 renv::init()
Now, in the shiny app directory, we can run
library(renv)
renv::init()For our package this returned the following:
> renv::init()
* Initializing project ...
* Discovering package dependencies ... Done!
* Copying packages into the cache ... Done!
The following package(s) will be updated in the lockfile:
# CRAN ===============================
- BH [* -> 1.72.0-3]
- R6 [* -> 2.4.1]
- Rcpp [* -> 1.0.5]
- assertthat [* -> 0.2.1]
- cli [* -> 2.0.2]
- crayon [* -> 1.3.4]
- digest [* -> 0.6.25]
- ellipsis [* -> 0.3.1]
- fansi [* -> 0.4.1]
- fastmap [* -> 1.0.1]
- fs [* -> 1.2.7]
- glue [* -> 1.4.2]
- htmltools [* -> 0.4.0]
- httpuv [* -> 1.5.2]
- jsonlite [* -> 1.6]
- later [* -> 1.0.0]
- lifecycle [* -> 0.2.0]
- magrittr [* -> 1.5]
- mime [* -> 0.6]
- pillar [* -> 1.4.6]
- pkgconfig [* -> 2.0.3]
- promises [* -> 1.1.0]
- renv [* -> 0.12.3]
- rlang [* -> 0.4.7]
- shiny [* -> 1.4.0]
- shinyFiles [* -> 0.8.0]
- sourcetools [* -> 0.1.7]
- tibble [* -> 3.0.3]
- utf8 [* -> 1.1.4]
- vctrs [* -> 0.3.4]
- xtable [* -> 1.8-3]
* Lockfile written to '~/ShinyApps/36-350_files/renv.lock'.
* Project '~/ShinyApps/36-350_files' loaded. [renv 0.12.3]
* renv activated -- please restart the R session.3.3.4 Package locations (not only CRAN)
This shiny app’s code is rather basic, but note that it captures dependences (like magrittr that I didn’t actually call in the app.). Additionally, it’s decided that all of these are from CRAN, but it can also use packages that are not in CRAN (see Package Sources for options).
3.3.5 How it stores this information
In the parent directory I also now have a file called renv.lock, this is a yaml file with all the package and R information, and looks like this:
{
"R": {
"Version": "3.6.0",
"Repositories": [
{
"Name": "CRAN",
"URL": "https://cloud.r-project.org"
}
]
},
"Packages": {
...,
"shiny": {
"Package": "shiny",
"Version": "1.4.0",
"Source": "Repository",
"Repository": "CRAN",
"Hash": "6ca23724bb2c804c1d0b3db4862a39c7"
},
...
}
3.4 installing and understanding the renv structure
You should also note that you can see that I have the shiny package, version 1.4.0 defined in this initialization. At the time of writing this, shiny, version 1.5.0 is on CRAN. Let’ suppose I wish to update, shiny, and run the following:
install.package("shiny")First, I’ll ok up an R console in this directly, and it will state:
... Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. * Project '~/ShinyApps/36-350_files' loaded. [renv 0.12.3]
This final line tells me I do have an renv available and active. I’ll run the installation file, and if I peek at the renv.lock file it still is telling me I care about shiny, version 1.4.0.
3.5 renv::status checking which differences exist
Like git I can check which packages differ in my renv compared to the current packages loaded with renv::status (and the following appears). This shows that withr, commonmark and base64enc are new packages loaded by the update and htmltools was also updated when I updated shiny.
> renv::status()
The following package(s) are installed but not recorded in the lockfile:
_
withr [2.3.0]
commonmark [1.7]
base64enc [0.1-3]
Use `renv::snapshot()` to add these packages to your lockfile.
The following package(s) are out of sync:
Package Lockfile Version Library Version
htmltools 0.4.0 0.5.0
shiny 1.4.0 1.5.0
Use `renv::snapshot()` to save the state of your library to the lockfile.
Use `renv::restore()` to restore your library from the lockfile.3.6 renv::restore
Suppose I really like version 1.4 instead of 1.5 for shiny (and my old snapshot). Then I can run:
renv::restore()This will ask me if I am sure I wish to revert, and then do so., getting me back to 1.4.0 for shiny:
> utils::packageVersion("shiny")
[1] ‘1.4.0’What’s nice about this is that - renv::restore should also restore my R package environment (aka renv :P ), if the server crashes and removes all my packages.
3.7 creating a new snapshot (renv::snapshot)
Let’s instead imagine I did actually want to snapshot/update shiny for my environment. Then (after reinstalling shiny 1.5.0), I can instead do
renv::snapshot()which will again ask if I’m sure and then update the renv.lock file to include my new packages (so if I run renv::status I’ll see the new package versions).
3.8 smart renv tools
Every time you reopen your directory in R it will do a check to tell you if you have any packages that don’t align with your renv - you can then correct that with a renv::restore() or renv::snapshot().
4. Other resources
Rstudioon Reprodicible Environments- Miles McBain on
drakeworkflow: probably more for data science pipelines than code you’ll work on in grad school - Miles McBain on motivation for
packrat(renvis the “2.0” ofpackrat)
<references />
