The {sandpaper} package was created by The Carpentries to re-imagine our method of creating lesson websites for our workshops. This package will take a series of Markdown or RMarkdown files and generate a static website with the features and styling of The Carpentries lessons including customized layouts and callout blocks. Much of the functionality in this package is inspired by Jenny Bryan’s work with the {usethis} package.


Please note: {sandpaper} is very much a work in progress. Please open an issue if you have any comments or suggestions!


Documentation

Want to know how this works in a lesson format? Head over to https://carpentries.github.io/sandpaper-docs/.

If, instead, you already know how a lesson is built and are interested in understanding how the functions in {sandpaper} work, you can visit this package documentation site at https://carpentries.github.io/sandpaper/.

Installation

{sandpaper} is not currently on CRAN, but it can be installed from our Carpentries Universe (updated every hour) with the following commands:

options(repos = c(
  carpentries = "https://carpentries.r-universe.dev/", 
  CRAN = "https://cran.rstudio.com/"
))
install.packages("sandpaper", dep = TRUE)

Note that this will also install development versions of the following packages:

package What it does
{varnish} html, css, and javascript templates for The Carpentries (in progress)
{tinkr} manipulation of knitr markdown documents built on the commonmark xml library
{pegboard} programmatic interface to lesson components for validation (in progress)

Design

This package is designed to make the life of the lesson contributors and maintainers easier by separating the tools needed to build the site from the user-defined content of the site itself. It will no longer rely on Jekyll or any of the other >450 static site generators, but instead rely on R, RStudio, and {pkgdown} to generate a site with the following features:

Rendering locally

In a repository generated via {sandpaper}, only the source is committed to avoid issues surrounding out-of-date artefacts and directory structure confusion.

The website is generated in two steps:

  1. markdown files from the source files are rendered containing a hash for the source file so that these need only be re-rendered when they change.
  2. html files are generated from the rendered markdown files and the CSS and JS sources in the {varnish} package for the preview.

To ensure there are no clashes between minor differences in the user setup, no artifacts are committed to the main branch of the repository. Because of the caching mechanism between the website and the rendered markdown files, long-running lessons can be updated and previewed quickly.

Rendering on continuous integration

Continuous integration will act as the single source-of-truth for how the outputs of the lessons are rendered. For this, we want the resulting website to be:

  • CI agnostic (but currently set up with GitHub)
  • easy to set up
  • auditable (e.g. I can see changes between the content of two commits)
  • versionable (e.g. I can instruct learners to go to <WEBSITE>/1.1. This is inspired from the python documentation style)

To acheive this, there will be two branches created: md-outputs and gh-pages that will inerit like so main -> md-outputs -> gh-pages. Because the build time from main to md-outputs can be time intensive, this will default to updating only files that were changed.

  • md-outputs: this branch will contain the files and artifacts generated from rmarkdown in the vignettes directory of a thin package skeleton.
  • gh-pages: this branch is generated via md-outputs and bundles the html, css, and js for the website. This will contain a single index.html file with several subfolders with different versions of the site. The index.html file will redirect to the current/ directory, which contains the up-to-date site.

Scheduled builds

  • gh-pages website: Because we are designing the lessons to have content separated from the styling, we will set up the CI to generate the webpage from the pre-built sources on a weekly basis, which will check if there has been an update to the styles (which I have in the {varnish} package) and then rebuild the site without rebuilding the content.
  • md-outputs branch: This will be rerun every month from scratch with the most recent version of R and R packages. If there is a change, a pull request can be generated to update the renv.lock file with a link to the changed markdown files in this branch.

Function syntax

The functions in {sandpaper} have the following prefixes:

  • create_ will create/amend files or folders in your workspace
  • update_ will update build resources in the lesson
  • build_ will build files from your source
  • check_ validates either the elements of the lesson and/or episodes
  • fetch_ will download files or resources from the internet
  • reset_ removes files or information
  • get_ will retrieve information from your source files as an R object
  • set_ will update information in files.
  • ci_ interacts with continous integration to build the website

Here is a working list of user-facing functions:

Lesson and Episode Creation

  • create_lesson() creates a lesson from scratch
  • create_episode() creates a new episode with the correct number prefix
  • create_dataset() creates a csv or text data set from an R object
  • set_episodes() arranges the episodes in a user-specified order

Accessors

Website Creation and Validation

  • check_lesson() checks and validates the source files and lesson structure
  • build_episode_md() renders an individual file to markdown (internal use)
  • build_episode_html() renders a built markdown file to html (internal use)
  • build_lesson() builds the lesson into a static website
  • build_portable_lesson() builds the lesson into a portable static website
  • fetch_lesson() fetches the static website from the lesson repository

Continuous Integration Utilities

  • ci_deploy() builds and deploys the lesson on CI from the source files
  • ci_build_markdown() builds the markdown files on CI from the source and deploys them to the markdown branch.
  • ci_build_site() deploys the lesson on CI from pre-rendered markdown files
  • ci_release() builds and deploys the lesson on CI from the source files and adds a release tag
  • update_github_workflows() updates GitHub workflows

Cleanup

Usage

There are five use-cases for {sandpaper}:

  1. Creating lessons
  2. Contributing to lessons
  3. Maintaining lessons
  4. Rendering a portable site
  5. Rendering a site with GitHub actions.

Creating a lesson

To create a lesson with {sandpaper}, use the create_lesson() function:

sandpaper::create_lesson("~/Desktop/r-intermediate-penguins")

This will create folder on your desktop called r-intermediate-penguins with the following structure:

|-- .gitignore               # - Ignore everything in the site/ folder
|-- .github/                 # - Scripts used for continuous integration
|   `-- workflows/           #
|       |-- deploy-site.yaml # -   Build the source files on github pages
|       |-- build-md.yaml    # -   Build the markdown files on github pages
|       `-- cron.yaml        # -   reset package cache and test
|-- episodes/                # - PUT YOUR MARKDOWN FILES IN THIS FOLDER
|   |-- data/                # -   Data for your lesson goes here
|   |-- figures/             # -   All static figures and diagrams are here
|   |-- files/               # -   Additional files (e.g. handouts) 
|   `-- 00-introducition.Rmd # -   Lessons start with a two-digit number
|-- instructors/             # - Information for Instructors
|-- learners/                # - Information for Learners
|   `-- setup.md             # -   setup instructions (REQUIRED)
|-- profiles/                # - Learner and/or Instructor Profiles
|-- site/                    # - This folder is where the rendered markdown files and static site will live
|   `-- README.md            # -   placeholder
|-- config.yaml              # - Use this to configure commonly used variables
|-- CONTRIBUTING.md          # - Carpentries Rules for Contributions (REQUIRED)
|-- CODE_OF_CONDUCT.md       # - Carpentries Code of Conduct (REQUIRED)
|-- LICENSE.md               # - Carpentries Licenses (REQUIRED)
`-- README.md                # - Introduces folks how to use this lesson and where they can find more information.

Once you have your site set up, you can add your RMarkdown files in the episodes folder. By default, they will be built in alphabetical order, but you can use the set_episodes() command to build the schedule in your config.yaml file:

s <- sandpaper::get_episodes()
sandpaper::set_episodes(order = s, write = TRUE)

When you want to preview your site, use the following:

sandpaper::build_lesson()

Working in RStudio?

If you are using RStudio, you can preview the lesson site using the keyboard shortcut ctrl + shift + B (which corresponds to the “Build Website” button in the “Build” tab. To preview individual files, you can use ctrl + shift + K (This corresponds to the “Knit” button in the editor pane)

This will create the website structure inside of the the site/ folder, render the RMarkdown files to markdown (for inspection and quick rendering), render the markdown files to HTML, and then enable a preview within your browser window.

Contributing to a Lesson

To contribute to a lesson, you can either fork the lesson to your own repository and clone it to your computer manually from GitHub, or you can use the {usethis} package to automate it. For example, This is how you can create a copy of Programming With R to your computer’s Desktop.

usethis::create_from_github(
  repo = "swcarpentry/r-novice-gapminder", 
  destdir = "~/Desktop/r-novice-gampinder",
  fork = TRUE
)

This will copy all of the source files to your computer and move you to the directory.

Note that the rendered website will not be immediately available. To download the site as it currently appears on the web, use:

sandpaper::fetch_lesson(markdown = TRUE, site = TRUE)

This will download the site and the rendered markdown files into the site/ folder. To save bandwidth, you can choose to just download the markdown files and artifacts by settin site = FALSE. Now, you can edit the Rmarkdown files in episodes/ and quickly render the site.

To upload changes to the lesson repository, you can use the follow

Maintaining a Lesson

When you are maintaining a lesson, there is a high likelihood that you will already have a copy on your machine. If not, follow the instructions in the contributing to a lesson section above.

The typical workflow will look like this:

  1. open the sandpaper project in RStudio and make edits to files in the episodes/ folder
  2. in the R console run the following
sandpaper::check_lesson() # validates the structure of the input files
sandpaper::build_lesson() # builds and validates lesson

Rendering a portable site

To render a portable site, you can follow the instructions for contributing to a lesson or maintaining a lesson to set up. Once you have the lesson set up, you can use the following command:

sandpaper::build_portable_lesson(version = "current")

This will render a fully portable lesson site as a zip file in the site/ folder. You can distribute this lesson to learners who do not have reliable internet access for use offline without sacrificing any of the styling.

Rendering with GitHub actions

Ultimately, there should be a minimal number of functions that handle this situation because writing CI configuration files is maddening. The most straightforward function is:

sandpaper::ci_deploy(md_branch = "md-outputs", site_branch = "gh-pages")

This function will create git worktrees for the orphan md-outputs branch in the site/built folder and the orphan gh-pages branch in the site/docs folder. After that, we generate the site as normal.

Because css and js libraries may need updating before any lesson material does, a step can be created just for rebuilding the site that uses:

sandpaper::ci_build_site(branch = "gh-pages")

When a lesson is given a release, the current site folder needs to be duplicated to a versioned folder and a tag needs to be added to the md-outputs branch:

sandpaper::ci_release(tag = "0.1", md_branch = "md-outputs", site_branch = "gh-pages")