CSS

Building My Site Part III: Fine Tuning

Final tweaks and performance improvements for my website.

TJ Fogarty

08 Jan 2018 • 3 min read

After making some decisions and implementing them, it’s now time to tidy a few things up and improve on performance. I’m going to talk about the ways in which I’ve improved the loading of fonts, CSS and JavaScript.

Fonts

I previously read an article called 23 Minutes of Work for Better Font Loading and it’s a brilliant piece of work outlining the ways in which font loading can be improved. I didn’t follow every step, but the two I implemented really made a difference:

First, preloading the web fonts by putting these tags into the <head> of my site:

<link rel="preload" href="/fonts/raleway.woff2" as="font" type="font/woff2" crossorigin>
<link rel="preload" href="/fonts/playfair.woff2" as="font" type="font/woff2" crossorigin>

Secondly, adding those fonts to my service worker using the SW Precache Webpack Plugin.

CSS

I opted to use the Tailwind CSS Framework to style my site. I found it a great way to throw a bunch of classes on my elements to rapidly style them, and abstract them out into their own classes once I was happy with them.

If you have a look around, you might notice that there isn’t a whole lot of style here. That may change in the future, but it also reminded me of Harry Roberts’ website where he inlined his styles within the <head> of his site rather than using a <link> tag.

Fair enough, but there’s a lot of CSS being pulled in with Tailwind that wouldn’t make sense to include. I’d only be needlessly increasing the weight of my page. Thankfully the author of this framework has a solution! PurgeCSS to the rescue:

// webpack.mix.js file
const mix = require('laravel-mix')
const tailwindcss = require('tailwindcss')
const glob = require('glob-all')
const PurgecssPlugin = require('purgecss-webpack-plugin')

// https://github.com/FullHuman/purgecss#extractor
class TailwindExtractor {
  static extract(content) {
    return content.match(/[A-z0-9-:\/]+/g)
  }
}

if (mix.inProduction()) {
  mix.webpackConfig(
      plugins: [
        new PurgecssPlugin({
          // Specify the locations of any files you want to scan for class names.
          paths: glob.sync([
            path.join(__dirname, 'templates/**/*.twig'),
            path.join(__dirname, 'web/assets/**/*.js')
          ]),
          extractors: [
            {
              extractor: TailwindExtractor,

              // Specify the file extensions to include when scanning for
              // class names.
              extensions: ['html', 'js', 'php', 'twig']
            }
          ]
        })
      ]
  )
}

This scans through template and JavaScript files, and strips out any classes that are not being used.

In layouts/default.twig I inline it using the Craft Mix plugin:

{{ mix('/assets/css/app.css', true, true) | raw }}

JavaScript

There’s some JavaScript on my site that isn’t required to be loaded with every single page, those being highlight.js and Algolia Search. Not every page requires syntax highlighting, and not everyone will click the search icon, so I needed a way to only load them when it was necessary. I did this with Dynamic Imports and some tweaking of my webpack.mix.js file.

When I was first using it, the chunks that were created were either dropped into the wrong directory, or the path they were loaded from were incorrect. Here’s what I added to my configuration to correct it:

mix.webpackConfig({
  output: {
    path: path.resolve(__dirname, 'web'),
    publicPath: '/',
    chunkFilename: 'assets/js/chunks/[name].js'
  }
})

Let’s look at syntax highlighting first. I wanted to check if a page had code on it, and load the library in if it did.

if (document.querySelector('pre')) {
    try {
      let hljs = await System.import(
        /* webpackChunkName: "hljs" */ 'highlight.js'
      )
      hljs.initHighlightingOnLoad()
    } catch (e) {
      console.log('Error loading highlight.js', e)
    }
}

Using the comment /* webpackChunkName: "hljs" */ I could specify the name of the generated file. Otherwise you’d end up with files called 0.js, 1.js etc…
Next up is the search. I’ve stripped out most of the interaction code here, and left in the loading of the required library:

import { env } from './utils'

export const Search = {
  trigger: document.querySelectorAll('.js-search'),
  index: null,

  init() {
    this.handleTriggerClick = this.handleTriggerClick.bind(this)

    this.trigger.forEach(trigger => {
      trigger.addEventListener('click', this.handleTriggerClick)
    })
  },

  handleTriggerClick(e) {
    e.preventDefault()
    this.loadSearchClient()
  },

  async loadSearchClient() {
    try {
      let algoliasearch = await System.import(
        /* webpackChunkName: "search" */ 'algoliasearch/lite'
      )

      let client = algoliasearch(
        'applicationId',
        'apiKey'
      )
      this.index = client.initIndex(
        env() === 'development' ? 'dev_posts' : 'prod_posts'
      )
    } catch (e) {
      console.log('Error loading search client', e)
    }
  }
}

I have an index for my local development version, and the production version. I have a function called env which helps me with which environment I’m in:

export const env = () => {
  return process && process.env && process.env.NODE_ENV
    ? process.env.NODE_ENV
    : null
}

Is it finished?

After all this, I have a deployment script that runs npm run production which minifies my assets, generates a service worker, and strips out unused CSS classes. Coupled with the font loading techniques I have a zippy little site.

There’s more I can do, however. Maybe there are smaller libraries out there that I can swap in, or some optimisations I can make to my artisanal, hand-rolled code. I could add the JavaScript to the service worker, though I have cache-invalidation trust issues that I need to work through first.

Until then, you can see the source code for this site as it stands today.