Spotlight

Hybrid rendering app for the GOV.UK Performance Platform using Backbone and D3. JavaScript is shared between the client and server, and the app makes use of progressive enhancement to provide a great experience in every browser.

Building and running the app

Development

Just Spotlight: The simplest way to get started is to run just this app, against production data.

Firstly, it is recommended that you set up Node Version Manager on your host. See the (nvm) README for installation instructions.

Next checkout the Spotlight repo and create an .nvmrc file in its root directory containing the version of node specified in the 'engines' entry in package.json e.g. 6.11.2.

Now install the specified version of node using nvm:

nvm install 6.11.2

To check you have the correct version of node installed:

nvm which

Found '/Users/<username>/<path to>/spotlight/.nvmrc' with version <6.11.2>
/Users/<username>/.nvm/versions/node/v6.11.2/bin/node

Now tell nvm to use the version of node specified in the .nvmrc file:

nvm use

You can then run the app as follows:

npm install
npm start

Now you should be able to connect to the app at http://localhost:3057.

The app uses node-supervisor and grunt-contrib-watch to monitor changes, automatically restart the server and recompile Sass.

By default, this will look at production data, but perhaps you want to connect to a different data source. You can do that by creating your own config file in /config/config.development_personal.json that mimics /config/config.development.json with a different backdropUrl property. It'll be ignored by Git.

Full stack: if you're using our development environment then you can run all our apps in one go and use a real database for development. As a bonus, this will let you test the image fallbacks using the screenshot-as-a-service app.

First, you need to set up the Performance Platform development environment.

Once you have a machine with the required system-level dependencies, you can run the application with:

cd /var/apps/pp-puppet/development
bowl performance

Running tests

Command line

The Jasmine tests are divided into ones that work on the client (test/spec/client), and ones that work on the server (test/spec/server and test/spec/shared). The client tests are run using Jasmine v2.x, while the server tests are using Jasmine v1.x. It used to be that both were written for Jasmine v1, but after upgrading node versions our client tests needed to be upgraded.

npm test runs both client and server tests, as well as linting the codebase:

npm run jasmine_node executes server Jasmine tests in Node.js
npm run jasmine executes client Jasmine tests in PhantomJS
npm run shell:cheapseats executes feature tests using cheapseats with a small subset of dashboards, for speed
npm run shell:cheapseats_full_run runs cheapseats with all dashboards
npm run test:functional executes functional tests using [nightwatch][https://github.com/beatfactor/nightwatch]

Functional tests

As part of the CI (travis) npm run test:functional:ci is run. This spins up an instance of spotlight, nightwatch and phantomjs to run the tests in a headless environment.

To assist with debugging the functional tests can also be run in a selenium webdriver using the following command npm run test:functional:ff

If you want to run against firefox,chrome and phantom you can also do npm run test:functional:all.

All the functional tasks except ci will require a server to be running already.

Debugging locally

Install node-inspector where the app runs with sudo npm install -g node-inspector@0.5.0 and run it with node-inspector.

Start the app with node --debug app/server.js and visit http://spotlight.perfplat.dev:8080/debug to view the console.

Production

npm run build:production to create a production release.

NODE_ENV=production node app/server.js to run the app in production mode.

Heroku

If you want to deploy the app to Heroku, follow these instructions.

Create an app on Heroku

Using the web interface, or the CLI:

heroku create <app-name>

Set the app to use the node-grunt buildpack

The app runs on Heroku using a custom buildpack for Grunt.js support.

This means it will run the grunt commands we need to compile the app when deploying code.

 heroku config:set BUILDPACK_URL=https://github.com/mbuchetics/heroku-buildpack-nodejs-grunt.git

Set configuration vars

heroku config:set NODE_ENV=development # makes app run in development mode
heroku config:set npm_config_production=true # does not install dev dependencies

Deploy the code

If the code you're deploying is not in master, then you'll need to make sure you specify your local branch to push to master. Otherwise it will just deploy your local master (and probably not work as expected).

git push heroku <your-branch-name>:master
heroku open # opens the freshly deployed app in a browser

Or just...

If you want the Heroku app to be password-protected, set config variables as follows, before pushing the code.

heroku config:set BASIC_AUTH_USER=xxxx
heroku config:set BASIC_AUTH_PASS=xxxx
heroku config

Logging

You might also want to enable some logging in your Heroku app to assist with debugging. You can use logentries to do that:

heroku addons:add logentries

You can then access the logs from your app's dashboard on Heroku (under the "Add-ons" section).

Contributing

For Javascript, follow the styleguide (apart from the sections on GOV.UK modules as we don't use these)

Functionality should work without Javascript where possible.

All content should work well with screenreaders (at least Voiceover and JAWS). 'Work well' means

a screenreader user can orientate themselves effectively and use the page.
async updates are reported to the user (an 'accessibility' module exists for this).

Notes for Developers

Tables

Tables are used in the following places:

To display a list of dashboards on the services, web-traffic and other dashboards pages
To display data on a dashboard, e.g. a web-traffic dashboard
To be displayed instead of a graph if javascript is disabled.

Configuration

As there are more tables underpinning graphs than any other tables in spotlight, table columns are configured as though they are the axes on a graph.

Columns defined in the x-axis will appear before those in the y-axis.

For example, axes defined as:

axes: {
    x: {
      key: 'key_X',
      label: 'Label X'
    },
    y: [
      {
        key: 'key_Y_one',
        label: 'Label Y1',
        format: 'integer'
      },
      {
        key: 'key_Y_two',
        label: 'Label Y2',
        format: 'integer'
      }
    ]
}

Will be displayed as:

Label X	Label Y1	Label Y2
Value X	Value Y One	Value Y Two

The y-axis only accepts a list of column configuration.

The x-axis will accept a list or a single value. If a list is provided, the columns will appear in reverse order in the table. For example, x-axis columns defined as:

x: [
  {
    key: 'key_X_one',
    label: 'Label X1'
  },
  {
    key: 'key_X_two',
    label: 'Label X2'
  }
]