IMPORTANT: While GitHub Actions look like JavaScript you could fork and try variants of your own, there is some overhead involved. See %BUILDING.md for a description of the necessary workflow.

Install and/or Invoke Ren-C With a GitHub Action

This is a JavaScript-based GitHub Action which will install a Ren-C executable appropriate for the platform of your GitHub container. It can also run some Ren-C code that you inline directly into your workflow YAML...just like shell code in the run: tag of a step.

The action can be run many times, but will only download the interpreter the first time. It uses the GitHub Actions tool-cache NPM package. Hence the installation will be under the $RUNNER_TOOL_CACHE environment variable, for instance /opt/hostedtoolcache/ren-c/3.0.0/x64/r3

While the executable is added to the path via GitHub Actions core.addPath() facility, you can also retrieve the exact path as one of the action outputs.

As an added capability, this same GitHub Action can use web automation of a local Firefox instance to run script code in the WebAssembly build. The "Marionette" driver is used to feed the code as keystrokes to the browser, running against hosted builds of the Web REPL.

Inputs

script

Optional script code to run in the interpreter. It accomplishes this by taking the code from the YAML and writing it to a file in the $RUNNER_TEMP directory and executing that.

(While this facility is currently modest and just runs the code as-is, there are ambitions for it to be used with configurations that would make a dialect context, so that one could immediately be jumped into a domain-specific tool but still using this same GitHub Action for installation and caching.)

commit

This is the GitHub short hash of the commit to use. It must be a deployed build (but not necessarily a "greenlit" build, which facilitates the usage of ren-c-action to do testing of a new build to see if it's worth of being marked as valid).

web (true or false, default is false)

Run script code in the online version of the interpreter, by means of browser automation. This feature currently only works on Ubuntu hosts, and uses the "Marionette" driver for controlling Firefox via WebSockets.

timeout (integer number of seconds)

A timeout can already be set on individual GitHub steps via timeout_minutes: However, interactions with the web build through the Marionette driver for Firefox offer a timeout facility in milliseconds. For the moment this is kept due to being more granular, but it only applies if web: true

checked (true or false, default is false)

Whether to use an instrumented build of the interpreter or not (e.g. a "debug" build). These are bigger and slower, but they help find bugs. You should run the checked build if at all possible, and any bug reports should be filed with what output the checked build gives on the situation.

Outputs

path

Full path where the r3 executable was installed.

Example usage

!!! This action is being developed right now and does not have a published version stamp yet. For the moment, the release branch is being used, and is often force-pushed. A formal v1.0 tag will be published as soon as that makes sense.

Plain installation, no script:

name: Install The Interpreter
uses: actions/ren-c-action@release

name: Demonstrate Usage 
run: |
   r3 --do "print {Executable is in the PATH}" 

Install and execute script code:

uses: actions/ren-c-action@release
with:
  script: |
    print {== Hello From R3 HTTPS Read Test! ==}
    parse as text! read https://example.com [
        thru <h1> copy header: to </h1> to end
    ] else [
        fail "Couldn't Capture Page Title"
    ]
    assert [header = "Example Domain"]
    print ["Succeeded:" header]

Install and retrieve path:

name: Install The Interpreter
id: install-r3
uses: actions/ren-c-action@release

name: Echo Installation Path
run: |
  echo "R3 Installed At ${{ steps.install-r3.outputs.path }}"