Skip to content

Latest commit

 

History

History
610 lines (427 loc) · 18.6 KB

README.md

File metadata and controls

610 lines (427 loc) · 18.6 KB

Runnable super-rentals tutorial

Build Status

What?

This allows you to create a runnable tutorial by embedding special runnable instructions (directives) in the markdown source, using the triple-backtick markdown code block syntax. These code blocks will be executed at build time and replaced with their output in the final markdown files.

Why?

  • Make sure the tutorial steps are up-to-date and working correctly with the latest Ember, Ember CLI EmberData, etc.
  • Save time by not having to manually sync the content with upstream blueprint changes!
  • Easy to maintain – changing a step early on in the tutorial automatically propagates to later steps (in code snippets, etc).
  • A fantastic integration test suite for the Ember ecosystem!

What?

run:command

Run one or more commands.

Example:

```run:command
ember new super-rentals
```

Result:

```
> ember new super-rentals
installing
  create .editorconfig
  create .ember-cli.js
  create .eslintignore
  create .eslintrc.js
  create .template-lintrc.js
  create .travis.yml
  create .watchmanconfig
  create README.md
  create app/app.js
  create app/index.html
  create app/resolver.js
  create app/router.js
  create app/styles/app.css
  create app/templates/application.hbs
  create config/environment.js
  create config/optional-features.json
  create config/targets.js
  create ember-cli-build.js
  create .gitignore
  create jsconfig.json
  create package.json
  create public/robots.txt
  create testem.js
  create tests/index.html
  create tests/test-helper.js
npm: Installed dependencies
Successfully initialized git.
```

The content of the source code block is the command(s) to run.

Commands can span multiple lines using \ at the end of each line to signal line-continuation, as in:

```run:command
echo "This is a \
  command that \
  spans multiple \
  lines."
```

Multiple commands can be supplied. If any of them fails, it will fail the build.

```run:command
npm run lint:hbs
npm run lint:js
npm run test
```

Lines starting with # and empty lines are ignored.

Options:

  • lang

    The syntax highlight language to use in the resulting code block. Defaults to shell.

  • hidden=true

    Run the command(s), but omit the code block from the final markdown file entirely.

  • cwd

    Specify a CWD (relative to dist/code) for the command. This defaults to . (i.e. dist/code), but most of the time you probably want to set it to super-rentals (i.e. dist/code/super-rentals). Unfortunately, we cannot just make that the default, because at the beginning of the tutorial, the folder does not exists yet. (Generating the app is part of the tutorial.)

  • captureCommand=false

    Run the command(s), but omit the command(s) themselves from the resulting code block.

  • captureOutput=false

    Run the command(s), but omit their output from the resulting code block.

run:file:create

Create a file.

Example:

```run:file:create lang=handlebars cwd=super-rentals filename=app/templates/index.hbs
<div class="jumbo">
  <div class="right tomster"></div>
  <h2>Welcome to Super Rentals!</h2>
  <p>We hope you find exactly what you're looking for in a place to stay.</p>
</div>
```

Result:

```handlebars { data-filename="app/templates/index.hbs" }
<div class="jumbo">
  <div class="right tomster"></div>
  <h2>Welcome to Super Rentals!</h2>
  <p>We hope you find exactly what you're looking for in a place to stay.</p>
</div>
```

The content of the source code block is used to populate the newly created file. It is also rendered into the resulting code block. A trailing newline will be added automatically, if it's not already included in the source code block.

Options:

  • lang

    The syntax highlight language to use in the resulting code block.

  • hidden=true

    Create the file, but omit the code block from the final markdown file entirely.

  • cwd

    Specify a CWD (relative to dist/code) for the filename. This defaults to . (i.e. dist/code), but most of the time you probably want to set it to super-rentals (i.e. dist/code/super-rentals). Otherwise, the resulting code block will have its data-filename set to super-rentals/app/..., which is probably not what you want. Unfortunately, we cannot just make that the default, because at the beginning of the tutorial, the folder does not exists yet. (Generating the app is part of the tutorial.)

  • filename (required)

    The filename (the path relative to cwd) used for creating the file. Also sets the data-filename metadata field in the resulting code block.

run:file:copy

Copy a file or folder from the assets folder.

Example:

```run:file:copy lang=css src=downloads/style.css cwd=super-rentals filename=app/styles/app.css
@import url(https://fonts.googleapis.com/css?family=Lato:300,300italic,400,700,700italic);

/**
 * Base Elements
 */

* {
  margin: 0;
  padding: 0;
}

/** ...snip... */
```

Result:

```css { data-filename="app/styles/app.css" }
@import url(https://fonts.googleapis.com/css?family=Lato:300,300italic,400,700,700italic);

/**
 * Base Elements
 */

* {
  margin: 0;
  padding: 0;
}

/** ...snip... */
```

If the source is a file, then the source file's content will be rendered into the resulting code block. If the source is a folder, its structure will be rendered into the resulting code block using a format similar to the Unix tree command.

If the source code block is non-empty, its content will be rendered into the resulting code block in place of the default output described above. This is useful because the file you are copying is probably quite large, and you don't necessarily want to render the whole file into the resulting markdown file.

Options:

  • lang

    The syntax highlight language to use in the resulting code block.

  • hidden=true

    Copy the file, but omit the code block from the final markdown file entirely.

  • cwd

    Specify a CWD (relative to dist/code) for the filename. This defaults to . (i.e. dist/code), but most of the time you probably want to set it to super-rentals (i.e. dist/code/super-rentals). Otherwise, the resulting code block will have its data-filename set to super-rentals/app/..., which is probably not what you want. Unfortunately, we cannot just make that the default, because at the beginning of the tutorial, the folder does not exists yet. (Generating the app is part of the tutorial.)

  • src (required)

    The source filename (the path relative to dist/assets) used for creating the file. Also sets the data-filename metadata field in the resulting code block.

  • filename (required)

    The filename (the path relative to cwd) used for creating the file. Also sets the data-filename metadata field in the resulting code block.

run:file:patch

Edit a file by applying a git patch.

Example:

```run:file:patch lang=js cwd=super-rentals filename=app/router.js
@@ -9,2 +9,3 @@
 Router.map(function() {
+  this.route('about');
 });
```

Result:

```js { data-filename="app/router.js" data-diff="+9" }
import EmberRouter from '@ember/routing/router';
import config from './config/environment';

const Router = EmberRouter.extend({
  location: config.locationType,
  rootURL: config.rootURL
});

Router.map(function() {
  this.router('about');
});

export default Router;
```

The content of the source code block is the git patch to apply.

A patch can be generated by modifying a file, and running git diff -U.

It is often a good idea to ask git to include minimal context to make the patch more resilient to changes in the blueprints. You can control the number of context lines included in the diff by passing a number to -U, such as git diff -U1. You can also manually edit and tweak the resulting patch to keep a useful amount of context for the task at hand.

It appears that the diff ... header line as well as the index ... line, as well as the "hunk context" (the text after the @@ ... @@) can be safely omitted. The --- filename and +++ filename lines are required by git, but can be omitted in the block; the directive will prepend them for you based on the filename argument if they are not already included in the patch.

A good workflow for generating patches:

  • Insert run:pause at the appropriate spot
  • Make sure the file you are editing is clean or staged (git add file)
  • Make the changes
  • git diff -U1 > diff.patch, play with the context number, tweak the patch by hand until you are happy with how it looks (keeping source-readability in mind)
  • Undo the changes with git checkout file
  • Test the patch with git apply diff.patch

Even though the patch contains line numbers, those are only used as "hints" when applying the diff. In practice, a well crafted patch could be quite resilient. For instance, the patch given in the example has been verified to apply cleanly even if the router blueprint has been changed to this:

import EmberRouter from '@ember/routing/router';
import config from './config/environment';

export default class Router extends EmberRouter {
 location = config.locationType;
 rootURL = config.rootURL;
}

Router.map(function() {
});

As you can see, even though the line numbers have shifted around, git has no trouble finding the relevant router map section from the above.

If the patch fails to apply cleanly, it will fail the build.

The resulting code block will contain the "combined" source of the file being edited, with data-diff metadata field indicated the removed and added lines. We can use this data on the client side to format the diff output. Potentially we can render it using an interactive component that allows you to toggle between the before/after/combined source, as well as folding away the unchanged lines.

Options:

  • lang

    The syntax highlight language to use in the resulting code block.

  • hidden=true

    Edit the file, but omit the code block from the final markdown file entirely.

  • cwd

    Specify a CWD (relative to dist/code) for the filename. This defaults to . (i.e. dist/code), but most of the time you probably want to set it to super-rentals (i.e. dist/code/super-rentals). Otherwise, the resulting code block will have its data-filename set to super-rentals/app/..., which is probably not what you want. Unfortunately, we cannot just make that the default, because at the beginning of the tutorial, the folder does not exists yet. (Generating the app is part of the tutorial.)

  • filename

    The filename (the path relative to cwd) used for creating the file. Also sets the data-filename metadata field in the resulting code block.

    This is also used to format the patch before sending it to git apply, so it is required unless they are already included in the patch and the block is set to hidden.

run:file:show

Render the content of a file or a folder.

Example:

```run:file:show lang=handlebars cwd=super-rentals filename=app/templates/index.hbs
```

Result:

```handlebars { data-filename="app/templates/index.hbs" }
<div class="jumbo">
  <div class="right tomster"></div>
  <h2>Welcome to Super Rentals!</h2>
  <p>We hope you find exactly what you're looking for in a place to stay.</p>
</div>
```

The content of the source code block is not used. If the source is a folder, its structure will be rendered into the resulting code block using a format similar to the Unix tree command.

Options:

  • lang

    The syntax highlight language to use in the resulting code block.

  • cwd

    Specify a CWD (relative to dist/code) for the filename. This defaults to . (i.e. dist/code), but most of the time you probably want to set it to super-rentals (i.e. dist/code/super-rentals). Otherwise, the resulting code block will have its data-filename set to super-rentals/app/..., which is probably not what you want. Unfortunately, we cannot just make that the default, because at the beginning of the tutorial, the folder does not exists yet. (Generating the app is part of the tutorial.)

  • filename (required)

    The filename (the path relative to cwd) used for reading the file. Also sets the data-filename metadata field in the resulting code block.

run:checkpoint

Indicates a checkpoint where the following steps are performed:

  • yarn test
  • Optionally, commit the current changes
  • Verify the git tree is clean (i.e. no dirty or untracked files)

This directive does not produce any output. If any of the steps failed, it will fail the build.

Example:

```run:checkpoint cwd=super-rentals
Chapter 1
```

The content of the source code block is the git commit message.

To avoid failing the "clean tree" test, you should be adding any created or modified files to the staging area as you go, using run:command hidden=true code blocks.

  • cwd

    Specify a CWD (relative to dist/code) for the command. This defaults to . (i.e. dist/code), but most of the time you probably want to set it to super-rentals (i.e. dist/code/super-rentals). Unfortunately, we cannot just make that the default, because at the beginning of the tutorial, the folder does not exists yet. (Generating the app is part of the tutorial.)

  • commit=false

    Don't create a git commit, but still run the other checks, including the "clean tree" test. This is only useful if the chapter did not make any changes at all, or one of the visible run:command blocks already committed the changes as part of the tutorial flow.

run:ignore (or run:ignore:*)

Ignore the source code block, and omit it from the final markdown file entirely.

This is useful for temporarily disabling a directive code block for debugging, or because it is not working, while still keeping the code in the source file. Essentially, this is how you "comment out" a directive code block.

For your convenience, you can pass any sub-directive after run:ignore:, or pass any arguments to it. This allows you to just insert :ignore: into an existing directive code block to disable it, without making any other changes.

Example:

```run:ignore:command cwd=super-rentals
# FIXME: don't run this for now, since Heroku is down atm
git push heroku main
```

run:pause

Pause the build until you are ready to resume.

This allows you to examine the state of things at a specific point in the tutorial, which is useful for debugging, taking screenshots or generating diff patches. Essentially, this is the this.pauseTest() for the tutorial.

Example:

```run:pause
Manually record a gif of performing the following steps:

...snip...
```

The content of the source code block will be printed to the command line prompt. This directive does not produce any output.

run:server:start

Start a server (background task).

Example:

```run:server:start cwd=super-rentals expect="Serving on http://localhost:4200/"
npm start
```

Result:

```shell
$ npm start
Build successful (9006ms) – Serving on http://localhost:4200/
```

The content of the source code block is the command for starting the server. Unlike run:command, you can only pass a single command, thought the command may span multiple lines if needed, using \ at the end of each line to signal line-continuation.

Lines starting with # and empty lines are ignored.

All servers started with this directive must be explicitly shut down with the run:server:stop directive before the end of the file, otherwise the build will fail.

Options:

  • id

    A unique identifier to reference this server process, which is needed when shutting it down later. This is optional; by default, the command to start the server is used as the id, but this allows you to specify a shorter name if desired.

  • lang

    The syntax highlight language to use in the resulting code block. Defaults to shell.

  • hidden=true

    Start the server, but omit the code block from the final markdown file entirely.

  • cwd

    Specify a CWD (relative to dist/code) for the command. This defaults to . (i.e. dist/code), but most of the time you probably want to set it to super-rentals (i.e. dist/code/super-rentals). Unfortunately, we cannot just make that the default, because at the beginning of the tutorial, the folder does not exists yet. (Generating the app is part of the tutorial.)

  • expect

    Wait for a particular string to appear on STDOUT to ensure the server has started successfully, before moving on to the next step.

  • timeout

    Wait for some time to pass (specified in seconds) before moving on to the next step.

    If used in conjunction with the expect option, it will fail the step if the checks are not completed before the deadline.

  • captureCommand=false

    Omit the command used to start the server from the resulting code block.

  • captureOutput=false

    Omit the output of the command used to start the server from the resulting code block.

run:server:stop

Stop a server (background task) previously started with run:server:start.

Example:

```run:server:stop
npm start
```

The content of the source code block is the command used to start the server. This directive does not produce any output.

Options:

  • id

    A unique identifier to reference the server process (see run:server:start). If this option is passed, the content block is ignored.

How?

  • Requires git
  • Probably only works on Unix/bash for now (PRs welcome)
    • Should probably run the build in a docker container anyway
  • yarn install
  • MAPBOX_ACCESS_TOKEN=your-token-here yarn build
    • Please note that you will need a Mapbox token in order to successfully run yarn build, otherwise the build will fail due to failing to load the map images. You can get your own token here. Once you have a token, you should assign it to the MAPBOX_ACCESS_TOKEN environment variable.
  • Processed markdown can be found in dist/markdown
  • The super-rentals code can be found in dist/code/super-rentals

Future Work

  • run:gif
  • Improve the build output
  • Extract this from super-rentals and make it usable for building arbitrary runnable tutorials

Prior Art