Merge branch 'main' into tooling-best-practices

building/config.json

@@ -836,6 +836,13 @@
     "title": "The Test Runner Interface",
     "blurb": ""
   },
+  {
+    "uuid": "0ab39b58-771e-42dc-aac9-e531bf43034e",
+    "slug": "tooling/test-runners/libraries",
+    "path": "building/tooling/test-runners/libraries.md",
+    "title": "Libraries",
+    "blurb": ""
+  },
   {
     "uuid": "a9cf4712-9022-4bd2-83aa-55cbdfd62374",
     "slug": "tooling/lines-of-code-counter",

building/tooling/test-runners/README.md

@@ -24,3 +24,4 @@ You can use the following documents to learn more about building a test runner:
 - [The Test Runner interface](/docs/building/tooling/test-runners/interface)
 - [How to build a Docker image with Docker for local testing and deployment](/docs/building/tooling/test-runners/docker)
 - [Best practices](/docs/building/tooling/best-practices)
+- [When and how to support libraries](/docs/building/tooling/test-runners/libraries)

building/tooling/test-runners/libraries.md

@@ -0,0 +1,33 @@
+# Libraries
+
+Exercism aims to teach [fluency](/docs/using/product/fluency) in the syntax, idioms, and the standard library of the language.
+This explicitly does _not_ mention external (non built-in) libraries, as that falls under _proficiency_.
+Exercises must thus, with few exceptions, be solvable without using libraries.
+
+## Reasons to support libraries
+
+There are couple of reasons why a track might still want to support libraries:
+
+1. The language has a (very) minimal standard library.
+   An example of such a language is Rust.
+2. The library adds testing functionality.
+   An example is a library that adds support for property-based testing.
+3. The exercise can only be solved using a library.
+   An example of such an exercise is the [lens-person exercise](https://exercism.org/exercises/lens-person), which in most languages can only be solved using a library.
+
+## Supporting libraries in the test runner
+
+As the test runner does not have access to the internet whilst running, it is not possible to download libraries at run time.
+The **only** solution to this problem is to install/download libraries at build time, where you _do_ have access to the internet.
+In practice, this means you'll need to install/download libraries within the Dockerfile.
+
+As an example, the [Prolog test runner's Dockerfile](https://github.com/exercism/prolog-test-runner/blob/ed7447a7518ede6ee3405e649f50aaec828e318b/Dockerfile) installs the `date_time` library:
+
+```dockerfile
+RUN swipl pack install date_time -y
+```
+
+## Documentation
+
+If your track supports libraries, this should be documented in a [track doc](/docs/building/tracks/docs).
+Please also link to this document (using its full URL) from the [`exercises/shared/.docs/help.md` document](/docs/building/tracks/shared-files#file-help-md).