Amy Kapernick sings us through numerous ways of improving the robustness and reliability of our front end code with a comprehensive rundown of ideas, tools, and resources. The girls and boys won’t get any toys until all the tests are passing.

Anyone who’s spoken to me at some point in November may get the impression that I’m a bit of a grinch. But don’t get me wrong, I love Christmas, I love decorating my tree, singing carols, and doing Christmas cooking - in December. So for me to willingly be humming the 12 days of Christmas in October, it’s probably for something that I think is even more important than banning premature Christmas decorations, like front end testing.

On the 12th day of Christmas, my front end dev, she gave to me, 12 testing tools, 11 optimised images, 10 linting rules, 9 semantic headings, 8 types of colour blindness, 7(.0) contrast ratio, 6 front end tests, 5 browser types, 4 types of tests, 3 shaken trees, 2 image types, and a source controlled deployment pipeline.

Twelve Testing Tools

1. axe does automated accessibility testing. Run as part of your development build, it outputs warnings to your console to let you know what changes you need to make (referencing accessibility guides). You can also specify particular accessibility standard levels that you’d like to test against, eg. best-practice, wcag2a or wcag2aa, or you can pick and choose individual rules that you want to check for (full list of rules you can test with axe).

2. BackstopJS runs visual regression tests on your website. Run separately, or as part of your deployment/PR process, you can use it to make sure your code changes aren’t bleeding into other areas of the website. By default, BackstopJS will set you up with a bunch of configuration options by running backstop init in your project to help get you started.

3. Website Speed Test analyses the performance of your website specifically with respect to images, and the potential size savings if they were optimised.

4. Calibre runs several different types of tests by leveraging Lighthouse. You can run it over your live website through their web app or through the command line, it then monitors your website for performance and accessibility over time, providing metrics and notifications of any changes.

5. Cypress is for end-to-end testing of your website. When visual regression testing may be a bit much for you, Cypress can help you test and make sure elements are still on the page and visible (even if they’re not pixel for pixel where they were last time).
6. pa11y is for automated accessibility testing. Run as part of your build process or using their CLI or dashboard, it tests your website against various Web Content Accessibility Guidelines (WCAG) criteria (including visual tests like colour contrast). While axe is run as part of your dev build and gives you an output to the console, it can be combined with pa11y to automate any changes as part of your build process.
7. whocanuse was created by Corey Ginnivan, and it allows you to view colour combinations as those with colour blindness would (as well as testing other visual deficiencies, and situational vision events), and test the colour contrast ratio based on those colours.

8. Colour Blindness Emulation was created by Kyo Nagashima as an SVG filter to emulate the different types of colour blindness, or if you’re using Gatsby, you can use a plugin based off of gatsby-plugin-colorblind-filters.
9. Accessible Brand Colors tests all your branding colours against each other (this is great to show designers what combinations they can safely use).

10. Browser dev tools - Most of the modern browsers have been working hard on the features available in their dev tools:
    • Firefox: Accessibility Inspector, Contrast Ratio testing, Performance monitoring.
    • Chromium: (Chrome, Edge Beta, Brave, Vivaldi, Opera, etc) - Accessibility Inspector, Contrast Ratio testing, Performance Monitoring, Lighthouse Audits (testing performance, best practices, accessibility and more).
    • Edge: Accessibility Inspector, Performance monitoring.
    • Safari: Accessibility Inspector, Performance monitoring.

11. Linc is a continuous delivery platform that makes testing the front end easier by automatically deploying a version of your website for every commit on every branch. One of the biggest hurdles when testing the front end is needing a live version of the site to view and test against. Linc makes sure you always have one.
12. ESLint and Stylelint check your code for programmatic and stylistic errors, as well as helping keep formatting standard on projects with multiple developers. Adding a linter to your project not only helps you write better code, it can reduce simple errors that might be found during testing time. If you’re not writing JavaScript, there are plenty of alternatives for whatever language you’re writing in.

If you’re trying to run eslint in VS Code, make sure you don’t have the Beautify extension installed, as that will break things.

Eleven Optimised Images

When it comes to performance, images are where we take the biggest hit, with images accounting for over 50% of total transfer size for websites. Many websites are serving excessively large images “just in case”, but there’s actually a native HTML element that allows us to serve different image sizes based on the screen size or serve better image formats when the browser supports it (or both).

<!-- Serving different images based on the width of the screen -->
<picture>
    <source
        srcset="/img/banner_desktop.jpg"
        media="(min-width: 1200px)"
    />
    <source
        srcset="/img/banner_tablet.jpg"
        media="(min-width: 700px)"
    />
    <source
        srcset="/img/banner_mobile.jpg"
        media="(min-width: 300px)"
    />
    <img src="/img/banner_fallback.jpg">
</picture>

<!-- Serving different image formats based on browser compatibility -->
<picture>
    <source
        srcset="/banner.webp"
        type="image/webp"
    />
    <img src="/img/banner_fallback.jpg">
</picture>

Ten Linting Rules

A year ago, I didn’t use linting. It was mostly just me working on projects, and I can code properly right? But these days it’s one of the first things I add to a project as it saves me so much time (and has taught me a few things about JavaScript). Linting is a very personal choice, but there are plenty of customisations to make sure it’s doing what you want, and it’s available in a wide variety of languages (including linting for styling).

// .eslintrc
module.exports = {
    rules: {
        'no-var': 'error',
        'no-unused-vars': 1,
        'arrow-spacing': ['error', { before: true, after: true }],
        indent: ['error', 'tab'],
        'comma-dangle': ['error', 'always'],
        // standard plugin - options
        'standard/object-curly-even-spacing': ['error', 'either'],
        'standard/array-bracket-even-spacing': ['error', 'either'], },
}

// .stylelintrc
{
    "rules": {
        "color-no-invalid-hex": true,
        "indentation": [
            "tab",
            {
                "except": [
                    "value"
                ]
            }
        ],
        "max-empty-lines": 2,
    }
}

Nine Semantic Headings

No, I’m not saying you should use 9 levels of headings, but your webpage should have an appropriate number of semantic headings. When your users are accessing your webpage with a screen reader, they rely on landmarks like headings to tell them about the page. Similarly to how we would scan a page visually, screen readers give users a list of all headings on a page to allow them to scan through the sections and access the information faster.

When there aren’t any headings on a page (or headings are being used for their formatting rather than their semantic meaning), it makes it more difficult for anyone using a screen reader to understand and navigate the page. Make sure that you don’t skip heading levels on your page, and remember, you can always change the formatting on a p tag if you need to have something that looks like a heading but isn’t one.

<h1>Heading 1 - Page Title</h2>
<p>Traditionally you'll only see one h1 per page as it's the main page title</p>
<h2>Heading 2</h2>
<p>h2 helps to define other sections within the page. h2 must follow h1, but you can also have h2 following another h2.</p>
<h3>Heading 3</h3>
<p>h3 is a sub-section of h2 and follows similar rules to h2. You can have a h3 after h3, but you can't go from h1 to h3.</p>
<h4>Heading 4</h4>
<p>h4 is a sub-section of h3. You get the pattern?</p>

Eight Types of Colour Blindness

Testing colour contrast may not always be enough, as everyone perceives colour differently. Take the below colour combination (ignoring the fact that it doesn’t actually look nice). It has decent colour contrast and would meet the WCAG colour contrast requirements for AA standards – but what if one of your users was red-green colour blind? Would they be able to tell the difference?

Red-green colour blindness is the most common form of colour blindness, but there are 8 different types affecting different parts of the colour spectrum, all the way up to complete colour blindness.

Protanopia

Inability to see red end of the colour spectrum.

Protanomaly

Difficulty seeing some shades of red.

Deuteranopia

Inability to see the green portion of the colour spectrum.

Deuteranomaly

Difficulty seeing some shades of green.

Tritanopia

Inability to see blue end of the colour spectrum.

Tritanomaly

Difficulty seeing some shades of blue.

Achromatopsia

Inability to see all parts of the colour spectrum, only able to perceive black, white and shades of grey.

Achromatomaly

Difficulty seeing all parts of the colour spectrum.

Seven (.0) Contrast Ratio

Sufficient colour contrast is perhaps one of the best steps to take for accessibility, as it benefits everyone. Having adequate contrast doesn’t just make the experience better for those with vision impairments, but it also helps those with situational impairments. Have you ever been in the sun and tried to read something on your screen? Whether you can view something when there’s glare could be as easy as making sure there’s enough contrast between the text and its background colour.

The WCAG have defined a contrast ratio of at least 4.5:1 for normal text (18.5px) and 3:1 for large text (24px) to meet AA accessibility standards, but this should be an absolute minimum and isn’t always readable. All four below examples have sufficient contrast to pass AA standards, but you might be hard pressed to read them when there’s glare or you have a dodgy monitor (even more so considering most websites use below 18.5px for their base font size).

To meet the AAA standard you need to have a ratio of 7:1 for normal text and 4.5:1 for large text, which should be sufficient for those with 20/80 vision to read.

Six Front End Tests

1. Adding default axe-core testing to Gatsby:

   //gatsby-config.js
   {
       resolve: 'gatsby-plugin-react-axe',
       options: {},
   },

2. Running pa11y tests on homepage at various screen sizes:

   // tests/basic-a11y_home.js
   const pa11y = require('pa11y'),
       fs = require('file-system')
   
   runTest()
   
   async function runTest() {
       try {
           const results = await Promise.all([
               pa11y('http://localhost:8000', {
                   standard: 'WCAG2AA',
                   actions: [],
                   screenCapture: `${__dirname}/results/basic-a11y_home_mobile.png`,
                   viewport: {
                       width: 320,
                       height: 480,
                       deviceScaleFactor: 2,
                       isMobile: true,
                   },
               }),
               pa11y('http://localhost:8000', {
                   standard: 'WCAG2AA',
                   actions: [],
                   screenCapture: `${__dirname}/results/basic-a11y_home_desktop.png`,
                   viewport: {
                       width: 1280,
                       height: 1024,
                       deviceScaleFactor: 1,
                       isMobile: false,
                   },
               }),
           ])
   
           fs.writeFile('tests/results/basic-a11y_home.json', JSON.stringify(results), err => {
               console.log(err)
           })
       } catch (err) {
           console.error(err.message)
       }
   }

3. Running pa11y tests on a blog post template at various screen sizes:

   // tests/basic-a11y_post.js
   const pa11y = require('pa11y'),
       fs = require('file-system')
   
   runTest()
   
   async function runTest() {
       try {
           const results = await Promise.all([
               pa11y('http://localhost:8000/template', {
                   standard: 'WCAG2AA',
                   actions: [],
                   screenCapture: `${__dirname}/results/basic-a11y_post_mobile.png`,
                   viewport: {
                       width: 320,
                       height: 480,
                       deviceScaleFactor: 2,
                       isMobile: true,
                   },
               }),
               pa11y('http://localhost:8000/template', {
                   standard: 'WCAG2AA',
                   actions: [],
                   screenCapture: `${__dirname}/results/basic-a11y_post_desktop.png`,
                   viewport: {
                       width: 1280,
                       height: 1024,
                       deviceScaleFactor: 1,
                       isMobile: false,
                   },
               }),
           ])
   
           fs.writeFile('tests/results/basic-a11y_post.json', JSON.stringify(results), err => {
               console.log(err)
           })
       } catch (err) {
           console.error(err.message)
       }
   }

4. Running BackstopJS on a homepage and blog post template at various screen sizes:

   // backstop.json
   {
     "id": "backstop_default",
     "viewports": [
       {
         "label": "phone",
         "width": 320,
         "height": 480
       },
       {
         "label": "tablet",
         "width": 1024,
         "height": 768
       },
       {
         "label": "desktop",
         "width": 1280,
         "height": 1024
       }
     ],
     "onBeforeScript": "puppet/onBefore.js",
     "onReadyScript": "puppet/onReady.js",
     "scenarios": [
       {
         "label": "Blog Homepage",
         "url": "http://localhost:8000",
         "delay": 2000,
         "postInteractionWait": 0,
         "expect": 0,
         "misMatchThreshold": 1,
         "requireSameDimensions": true
       },
       {
         "label": "Blog Post",
         "url": "http://localhost:8000/template",
         "delay": 2000,
         "postInteractionWait": 0,
         "expect": 0,
         "misMatchThreshold": 1,
         "requireSameDimensions": true
       }
     ],
     "paths": {
       "bitmaps_reference": "backstop_data/bitmaps_reference",
       "bitmaps_test": "backstop_data/bitmaps_test",
       "engine_scripts": "backstop_data/engine_scripts",
       "html_report": "backstop_data/html_report",
       "ci_report": "backstop_data/ci_report"
     },
     "report": [
       "browser"
     ],
     "engine": "puppeteer",
     "engineOptions": {
       "args": [
         "--no-sandbox"
       ]
     },
     "asyncCaptureLimit": 5,
     "asyncCompareLimit": 50,
     "debug": false,
     "debugWindow": false
   }

5. Running Cypress tests on the homepage:

   // cypress/integration/basic-test_home.js
   describe('Blog Homepage', () => {
       beforeEach(() => {
           cy.visit('http://localhost:8000')
       })
       it('contains "Amy Goes to Perth" in the title', () => {
           cy.title().should('contain', 'Amy Goes to Perth')
       })
       it('contains posts in feed', () => {
           cy.get('.article-feed').find('article')
       })
       it('all posts contain title', () => {
           cy.get('.article-feed')
               .find('article')
               .get('h2')
       })
   })

6. Running Cypress tests on a blog post template at various screen sizes:

   // cypress/integration/basic-test_post.js
   
   describe('Blog Post Template', () => {
       beforeEach(() => {
           cy.visit('http://localhost:8000/template')
       })
       it('contains "Amy Goes to Perth" in the title', () => {
           cy.title().should('contain', 'Amy Goes to Perth')
       })
       it('has visible post title', () => {
           cy.get('h1').should('be.visible')
       })
       it('has share icons', () => {
           cy.get('.share-icons a').should('be.visible')
       })
       it('has working share icons', () => {
           cy.get('.share-icons a').click({ multiple: true })
       })
       it('has a visible author profile image', () => {
           cy.get('.author img').should('be.visible')
       })
   })
   
   describe('Mobile Blog Post Template', () => {
       beforeEach(() => {
           cy.viewport('samsung-s10')
           cy.visit('http://localhost:8000/template')
       })
       it('contains "Amy Goes to Perth" in the title', () => {
           cy.title().should('contain', 'Amy Goes to Perth')
       })
       it('has visible post title', () => {
           cy.get('h1').should('be.visible')
       })
       it('has share icons', () => {
           cy.get('.share-icons .share-link').should('be.visible')
       })
       it('has a visible author profile image', () => {
           cy.get('.author img').should('be.visible')
       })
   })

Five Browser Types

Browser testing may be the bane of our existence, but it’s gotten easier, especially when you know the secret:

Not every browser needs to look the same.

Now, this may differ depending on your circumstances, but your website doesn’t have to match pixel for pixel across all browsers. As long as it’s on-brand and is useable across all browsers (this is where a good solid HTML foundation is useful), it’s ok for your site to look a little different between browsers.

While the browsers you test in will differ depending on your user base, the main ones you want to be covering are:

• Chrome/Chromium
• Firefox
• Safari
• Internet Explorer
• Edge

Make sure you’re testing these browsers on both desktop and mobile/tablet as well, sometimes their level of support or rendering engine will differ between devices – for example, iOS Chrome uses the Safari rendering engine, so something that works on Android Chrome may not work on iOS Chrome.

Four Types of Test

When it comes to testing the front end, there are a few different areas that we can cover:

1. Accessibility Testing: doing accessibility testing properly usually involves getting an expert to run through your website, but there are several automated tests that you can run against various standard levels.
2. Performance Testing: performance testing does technically bleed into the back end as well, but there are plenty of things that can be done from a front end perspective. Making sure the images are optimised, our code is clean and minified, and even optimising fonts using features like the font-display property. No amount of optimising the server and back end will matter if it takes forever for the front end to appear in a browser.
3. Visual Regression Testing: we’ve all been in the position where changing one line of CSS somewhere has affected another section of the website. Visual regression testing helps prevent that. By using a tool that compares before and after screenshots against one another to flag up what’s changed, you can be sure that style changes won’t bleed into unintended areas of the site.
4. Browser/device testing: while we all want our users to be running the most recent version of Chrome or Firefox, they may still be using the inbuilt browser on their DVD player – so we need to test various browsers, platforms and devices to make sure that our website can be accessed on whatever device they use.

Three Shaken Trees

Including (and therefore requiring your users to download) things that you’re not using affects the performance of your application. Are you forcing them to download the entire lodash library when you’re only using 2 functions? While a couple of unused lines of code may not seem like a huge performance hit, it can greatly affect users with slower devices or internet connections, as well as cluttering up your code with unused functions and dependencies. This can be set up on your bundler – Webpack and Parcel both have guides for tree shaking, and Gatsby has a plugin to enable it.

Two Image Types

While there are several different types of images, most of the time they fall into one of two categories:

Informative

The image represents/conveys important information that isn’t conveyed by the content surrounding it.

Decorative

The image only adds visual decoration to a page.

From these two categories, we can then determine if we need to provide alternative text for an image. If an image is purely decorative, then we add alt="" to let screen readers know that it’s not important. But if an image is informative, then we need to be supplying a text alternative that describes the picture for anyone who’s using a screen reader or isn’t able to see the image (remember the days when a standard internet connection took a long time to load a page and you saw alt text before an image loaded).

<img src="./nice-picture.jpg" alt="" />
<img src="./important-graphic.png" alt="This is a picture of something important to help add meaning to the text around me" />

If you have a lot of images with missing alt text, look into services that can auto-generate alt text based on image recognition services.

One Source Controlled Deployment Pipeline

While front end tests are harder to automate, running them through a source control and deployment pipeline helps track changes and eliminates issues where “it works on my computer”. Whether you’re running tests as part of the PR process, or simply against every commit that comes through, running tests automatically as part of your process makes every developer’s life easier and helps keep code quality at a high standard.

We already knew that testing was important, and your project can’t be run unless all your unit and integration tests are written (and pass), but often we forget about testing the front end. There are so many different tests we need to be running on the front end, it’s hard to work out what your need to test for and where to start.

Hopefully this has given you a bit of insight to front end testing, and some Christmas cheer to take you into the holidays.

About the author

Amy wears many hats as a freelance developer, business owner and conference addict. She regularly shares her knowledge with her peers and the next generation of developers by mentoring, coaching, teaching and feeding into the tech community in many ways.

Amy can be found volunteering her time with Fenders, ACS, SheCodes (formerly Perth Web Girls) and MusesJS (formerly NodeGirls). She also works as an evangelist for YOW! Conferences, is a Twilio Champion and has been nominated for the WiTWA awards for the last 2 years.

In her spare time Amy shares her knowledge and experience on her blogs and speaking at conferences. She has previously given keynotes at multiple events as well as speaking at several international conferences in the US and Europe.

More articles by Amy

Jason Pamental forges a path through the freshly laid snowy landscape of variable fonts. Like a brave explorer in a strange new typography topology let Jason show you the route to some fantastic font feats. Everything you thought you knew has changed.

Everything you thought you knew about fonts just changed (for the better).

Typography has always been a keen interest of mine, long before we were able to use fonts on the web. And while we’ve had the ability to that now for ten years, we’ve always been constrained by balancing the number of fonts we want to use with the amount of data to be downloaded by the viewer. While good type and typography can bring huge benefits to design, readability, and overall experience—include too many fonts and you negatively impact performance and by extension, user experience. Three years ago, an evolution of the OpenType font format was introduced that changes things in some really remarkable ways.

Introducing OpenType Font Variations (aka ‘variable fonts’)

As long as I’ve used digital fonts, I’ve had to install separate files for every width, weight, or variant that I want to use. Bold in one file, light in another, condensed italic another one yet again. Installing a whole family for desktop use might involve nearly 100 files. The variable font format is an evolution of OpenType (the format we’ve all been using for years) that allows a single file to contain all of those previously separate files in a single, highly efficient one. The type designer can decide which axes to include, and define minimum and maximum values.

On the web, that means we can load a single file and use CSS to set any axis, anywhere along the allowable range, without any artificial distortion by the browser. Some fonts might only have one axis (weight being the most common), and some may have more. A few are defined as ‘registered’ axes, which are the most common: width, weight, slant, italic, and optical size—but the format is extensible expressly so that designers can define their own custom axes and allow any sort of variation they want to create. Let’s see how that works on the desktop.

Just like before, but different

One of the ways the new format preserves backwards compatibility with other applications that don’t yet explicitly support variable fonts is something called ’named instances’—which are essentially mapped aliases for what used to be separate files. So whatever the typeface designer had in mind for ‘bold condensed’ would simply map to the appropriate points on the variation axes for weight and width. If the font has been made correctly, those instances will allow the font to be installed and used in recent versions of Windows and the MacOS just like they always have been.

If the application fully supports variable fonts, then you would also be able to manipulate individual axes as you see fit. Currently that includes recent versions of Adobe Illustrator, Photoshop, and InDesign, and also recent versions of the popular web/UI design application Sketch.

Discovering the secrets of style

To get all of the specifics of what a font supports, especially for use on the web, you’ll want to do one of two things: check the following website, or download Firefox (or better, do both).

If you have the font file and access to the web, go check out Roel Nieskens’ WakamaiFondue.com (What Can My Font Do… get it?). Simply drag-and-drop your font file as directed, and you’ll get a report generated right there showing what features the font has, languages its supports, file size, number of glyphs, and all of the variable axes that font supports, with low/high/default values displayed. You even get a type tester and some sliders to let you play around with the different axes. Take note of the axes, values, and defaults. We’ll need that info as we get into writing our CSS.

If you don’t have access to the font file (if it’s hosted elsewhere, for example), you can still get the information you need simply by using it on a web page and inspecting it with the Firefox developer tools. There are lots of fantastic videos on them (like this one and this one), but here’s the short version.

Thanks to Jen Simmons and the FF dev tools team, we have some incredible tools to work with web fonts right in the browser. Inspect a text element in the font you’re looking to use, and then click on the ‘fonts’ tab over to the right. You’ll then be greeted with a panel of information that shows you everything about the font, size, style, and variation axes right there! You can even change any of those values and see it rendered right in the browser, and if you then click on the ‘changes’ tab, you can easily copy and paste the changed CSS to bring right back into your code.

Now that you have all of the available axes, values, defaults, and their corresponding 4-character axis ’tags’—let’s take a look at how to use this information in practice. The first thing to note is that the five ‘registered’ axes have lower-case tags (wght, wdth, ital, slnt, opsz), whereas custom axis tags are always uppercase. Browsers are taking note, and mismatching upper and lower case can lead to unpredictable results.

There are two ways to implement the registered axes: through their corresponding standard CSS attributes, and via a lower-level syntax of font-variation-settings. It’s very important to use the standard attributes wherever possible, as this is the only way for the browser to know what to do if for some reason the variable font does not load, or for any alternate browsing method to infer any kind of semantics from our CSS (i.e. a heavier font-weight value signifying bolder text). While font-variation-settings is exactly what we should be using for custom axes (and for now, with italics or italics and slant axes), font-weight (wght) and font-stretch (wdth) are both supported fully in every browser that supports variable fonts. Now let’s have a look at the five registered axes and how to use them.

Weight

Probably the most obvious axis is weight—since almost every typeface is designed with at least regular and bold weights, and quite often much lighter/thinner and bolder extremes. With a variable font, you can use the standard attribute of font-weight and supply a number somewhere between the minimum and maximum value defined for the font rather than just a keyword like normal or bold. According to the OpenType specification, 400 should equate to normal for any given font, but in practice you’ll see that at the moment it can be quite varied by typeface.

p {
  font-weight: 425;
}
strong {
  font-weight: 675;
}

Why you’ll like this

Besides being able to make use of a broader range for things like big quotes in an extra-thin weight, or adding even more emphasis with a super-chonky one, you should try varying what it means for something to be ‘bold’. Using a ’slightly less bold’ value for bold text inline with body copy (i.e. the ’strong’ tag) can bring a bit more legibility to your text while still standing out. The heavier the weight, the more closed the letterforms will be, so by getting a bit more subtle at smaller sizes you can still gain emphasis while maintaining a bit more open feel. Try setting strong to a font-weight somewhere between 500-600 instead of the default 700.

Width

Another common variation in typeface design is width. It’s often seen referred to as ‘condensed’ or ‘compressed’ or ‘extended’—though the specifics of what these keywords mean is entirely subjective. According to the spec, 100 should equate to a ’normal’ width, and valid values can range from 1 to 1000. Like weight, it does map to an existing CSS attribute—in this case the unfortunately-named font-stretch attribute and is expressed as a percentage. In these early stages of adoption many type designers and foundries have not necessarily adhered to this standard with the numeric ranges, so it can look a little odd in your CSS. But a width range of 3%-5% is still valid, even if in this case 5% is actually the normal width. I’m hopeful that with more nudging we’ll see more standardization emerge.

p {
  font-stretch: 89%;
}

Why you’ll like this

One of the tricky things about responsive design is making sure your larger headings don’t end up as monstrous one-word-per-line ordeals on small screens. Besides tweaking font-size, try making your headings slightly narrower as well. You’ll fit more words per line without sacrificing emphasis or hierarchy by having to make the font-size even smaller.

Italic

The Italic axis is more or less what you’d expect. In most cases it’s a boolean 0 or 1: off (or upright) or on—usually meaning slanted strokes and often glyph replacements. Often times the lower case ‘a’ or ‘g’ have slightly different Italic forms. While it’s certainly possible to have a range rather than strictly 0 or 1, the off/on scenario is likely the most common that you’ll encounter. Unfortunately, while it is intended to map to font-style: italic, this is one of the areas where browsers have not fully resolved the implementation so we’re left having to rely upon the lower-level syntax of font-variation-settings. You might give some thought to using this in conjunction with a CSS custom property, or variable, so you don’t have to redeclare the whole string if you just want to alter the Italic/upright specification.

:root {
  --text-ital: 0;
}
body {
  font-variation-settings: 'ital' var(--text-ital);
}
em {
   --text-ital: 1;
}

Why you’ll like this

Having Italics as well as upright, along with weight and any other axes available, means you can use one or two files instead of 4 to handle your body copy. And with the range of axes available, you might just not need anything else.

Slant

The slant axis is similar to Italic, but different in two key ways. First, it is expressed as a degree range, and according to the OpenType specification should be ‘greater than -90 and less than +90’, and second, does not include glyph substitution. Usually associated with sans-serif typeface designs, it allows for any value along the range specified. If the font you’re using only has a slant axis and no italics (I’ll talk about that in a bit), you can use the standard attribute of ‘font-style’ like so:

em {
   font-style: oblique 12deg;
}

If you have both axes, you’ll need to use font-variation-settings—though in this case you just supply a numeric value without the deg.

:root {
  --text-slnt: 0;
}
body {
  font-variation-settings: 'slnt' var(--text-slnt);
}
em {
   --text-slnt: 12;
}

Why you’ll like this

The slant axis allows for anything within the defined range, so opportunities abound to set the angle a little differently, or add animation so that the text becomes italic just a little after the page loads. It’s a nice way to draw attention to a text element on the screen in a very subtle way.

Optical Size

This is a real gem. This is a practice that dates back over 400 years, whereby physically smaller type would be cut with slightly thicker strokes and a bit less contrast in order to ensure they would print well and be legible at smaller sizes. Other aspects can be tailored as well, like apertures being wider, terminals more angled, or bowls enlarged. Conversely, larger point sizes would be cut with greater delicacy, allowing for greater contrast and fine details. While this was in many ways due to poorer quality ink, paper, and type—it still had the effect of allowing a single typeface design to work optimally at a range of physical sizes. This practice was lost, however, with the shift to photo typesetting and then digital type. Both newer practices would take a single outline and scale it, so either the fine details would be lost for all, or the smaller sizes would end up getting spindly and frail (especially on early lower-resolution screens). Regaining this technique in the form of a variable axis gives tremendous range back to individual designs.

The concept is that the numeric value for this axis should match the rendered font-size, and a new attribute was introduced to go along with it: font-optical-sizing. The default is auto, and this is supported behavior in all shipping browsers (well, as soon as Chrome 79 ships). You can force it to off, or you can set an explicit value via font-variation-settings.

body {
  font-optical-sizing: auto;
}

Or:

:root {
  --text-opsz: 16;
}
body {
  font-variation-settings: 'opsz' var(--text-opsz);
}
h1 {
   --text-opsz: 48;
  font-size: 3em;
}

Why you’ll like this

A good optical size axis makes type more legible at smaller sizes, and tailoring that to the size it’s used makes a remarkable difference. On the other end of the spectrum, the increased stroke contrast (and anything else the type designer decides to vary) can mean a single font can feel completely different when used larger for headings compared with body copy. Look no further than Roslindale from David Jonathan Ross’ Font of the Month Club, in use on my site to see how big a difference it can be. I’m using a single font for all the headings and body copy, and they feel completely different.

Slant & Italics

I’m not sure that the creators of the specification were thinking of this when it was written, but technically there is no reason you can’t have separate axes for slant (i.e. angle) and Italic (i.e. glyph substitution). And indeed both DJR and Stephen Nixon have done just that, with Roslindale Italic and Recursive, respectively. With Recursive, you can see how much greater flexibility you can get by separating the angle from the glyphs. It can impart a completely different feel to a block of text to have an angle without the alternate forms. With the state of Italic implementation and the fact that they share the same CSS attribute, this is on that requires the use of font-variation-settings in order to set the attributes separately.

:root {
  --text-ital: 0;
  --text-slnt: 0;
}
body {
  font-variation-settings: 'ital' var(--text-ital), 'slnt' var(--text-slnt);
}
em {
   --text-ital: 1;
   --text-slnt: 12;
}
.slanted {
   --text-slnt: 12;
}
.italic-forms-only {
  --text-ital: 1;
}

Why you’ll like this

Having these axes separated can give you greater design flexibility when creating your typographic system. In some cases you might opt for a slant only, in others both angle and glyph substitution. While it may not be the most critical of features, it does add an extra dimension to the utility and dynamic range of a font.

Custom axes

While so far there are only five ‘registered’ axes, type designers can also create their own. Any aspect of the typeface design could potentially become an axis. There are the more ‘expected’ ones like serif shape or perhaps x-height (the height of the lower case letters) to much more inventive ones like ‘gravity’ or ‘yeast’. I’ll let someone else elaborate on those, but I will show an example of one I hope will become more common in text and UI designs: grade.

Grade

The notion of ‘grade’ in a typeface was first introduced to compensate for ink gain on different kinds of paper and presses as a way to visually correct across workflows and have a typeface appear the same on every one. The concept is that you’re essentially altering the weight of the font without changing the spacing. Having this as a variable axis can be useful in a couple of ways. Creating a higher-contrast mode, where the text gets a bit heavier without reflowing, can make text more legible in lower-light situations or in designing for ‘dark mode’. And when animating interface elements it can be add a bit heavier text grade along with a background color shift on hover or tap. It can also come in handy in responding to lower-resolution screens, where type can easily become a bit spindly. Note that custom axes need to be specified in all caps.

:root {
  --text-GRAD: 0;
}
body {
  font-variation-settings: 'GRAD' var(--text-GRAD);
}
body.dark {
  --text-GRAD: 0.5;
}

Why you’ll like this

I think the biggest use for a grade axis will be for accessibility—designing for things like a dark or high-contrast mode. But you’ll also be able to have some fun with UI animations, like making text heavier on buttons or navigation on hover or focus without altering the physical space occupied by the text.

Support

Happily support for variable fonts is quite good: recent versions of MacOS and Windows offer support at the OS level, meaning that they can be installed on your system and if the font has any ’named instances’, they will show up in any application’s font menu just as if they were separate fonts. If you have recent versions of Adobe CC applications Illustrator, Photoshop, or InDesign—or recent versions of Sketch—you can manipulate all of the available axes. In browsers, it’s better, and has been for quite some time. According to CanIUse.com it’s around 87%, but the more relevant bit for most is that both dominant mobile platforms and all the major shipping browsers support them.

The only really glaring exception is IE11, and given that you can easily use @supports to scope the inclusion of variable fonts it’s perfectly safe to put them in production today. That’s the strategy in use on the new web platform for the State of Georgia in the US, and it’s been deployed on over 40 sites so far and is happily serving static fonts to state employees (IE11 is their default browser) and variable ones to millions of citizens across the state.

p {
  font-family: YourStaticFontFamily;
}
@supports (font-variation-settings: normal) {
  p {
    font-family: YourVariableFontFamily;
  }
}

Since CSS is always parsed completely before any other action is taken, you can be sure that browsers will never download both assets.

Getting the fonts in your project

For now, many of you will likely be self-hosting your variable fonts as at this point only Google is offering them through their API, and so far only in beta. There are a few key differences in how you structure your @font-face declaration, so let’s have a look.

@font-face {
  font-family: "Family Name";
  src: url("YourVariableFontName.woff2")
    format("woff2 supports variations"), url("YourVariableFontName.woff2")
    format("woff2-variations");
  font-weight: [low] [high];
  font-stretch: [low]% [high]%;
  font-style: oblique [low]deg [high]deg;
}

The first thing you might notice is that the src line is a bit different. I’ve included two syntaxes pointing to the same file, because the official specification has changed, but browsers haven’t caught up yet. Because we have color fonts on the horizon in addition to variable ones (and the possibility that some may be both variable and in color), the syntax needed to be more flexible. Thus the first entry—which could specify ‘woff2 supports variations color’ for a font that supports both. Once browsers understand that syntax, they’ll stop parsing the ’src’ line once they get here. For now, they’ll skip that and hit the second one with a format of woff2-variations, which all current browsers that support variable fonts will understand.

For weight (font-weight) and width (font-stretch), if there is a corresponding axis, supply the low and high values (with the percentage symbol for width values). If there is no corresponding axis, just use the keyword ‘normal’. If there is a slant axis, supply the low and high values with ‘deg’ after each number. It’s worth noting that if there is also an italic axis (or only an italic axis and no slant), it’s best at this point to simply omit the font-style line entirely.

By supplying these values, you create some guard rails that will help the browser know what to do if the CSS asks for a value outside the allowed range. This way if the weight range is 300-700 and you accidentally specify font-weight: 100, the browser will simply clamp to 300 and won’t try to synthesize a lighter weight. It’s worth noting that this only works with the standard CSS attributes like font-weight or font-stretch. If you use font-variation-settings to set values, the browser assumes you’re the expert and will attempt to synthesize the result even if it’s outside the normal range.

Google Fonts is on the case, too

Back in September, the Google Fonts team announced a beta version of their API that supports some variable fonts. That support is growing, and more fonts are on the way. If you want to play around with it today though, you can have a look at an article I wrote about how, and check out a CodePen I created that’s using it.

Where to find them

The first place you should look for variable fonts is Nick Sherman’s v-fonts.com, which has been serving as a de facto catalog site, listing pretty much every variable font available. You can also have a look on GitHub where you’ll find a bunch of projects (in varying stages of completeness, but there are some good ones to be found). Nick also maintains a Twitter account that will tweet/retweet lots of announcements and links, and I publish a newsletter on web typography where I’ll generally include a few links to noteworthy releases.

You can also check out Laurence Penney’s Axis-Praxis.org site, the original variable fonts playground where you can put many of them (or even upload your own) into a type testing page that can give you loads of additional detail about available font features.

In truth, many designers and foundries are experimenting with making them, so if you’re unsure about availability it’s always worthwhile to ask. Get in touch and I can probably help make the connection!

Why it all matters

While all of this might be interesting purely from an academic standpoint, there are some significant benefits and opportunities that come from adopting variable fonts. From a performance standpoint, while variable fonts may be larger than single-instance font files, they are still far smaller than the sum total of static files they replace—and often come in smaller than 3-4 single fonts. Which means that page load times could substantially improve. This is the driving motivation for Nielson/Norman Group’s inclusion of Source Sans Variable on their site last year, or what Google has been testing with Oswald Variable on sites 148 million times a day for the past several months. Basically just using them instead of a few static instances to reap the benefits of faster page loads and less code.

But beyond that, what really excites me are the design possibilities. Once we have variable fonts on our sites, we’re free to get infinitely more expressive. And with the sophistication of our publishing systems, building some of that flexibility into our publishing process should not be far behind. So creating things like my experiment below shouldn’t be one-off exceptions, but rather part of a regular practice of bringing design back into the publishing process.

Go have fun

I hope this has served as a good starting point to get into designing and developing with variable fonts. Send links and questions—I can’t wait to see what you make! And stay tuned—there just might be another post coming that goes even further ;)

In the meantime, if you want to learn more about integrating variable fonts with all sorts of other ideas, check out the ever-amazing Mandy Michael’s site variablefonts.dev.

About the author

Jason spends much of his time working with clients to establish their typographic systems and digital strategy, helping design and development teams works smarter and faster, and running workshops about all of the above. He is a seasoned design and user experience strategy leader with over 20 years’ experience on the web in both creative and technical roles, and an Invited Expert to the W3C Web Fonts Working Group. Clients range from type industry giants, Ivy League and High Tech, to the NFL and America’s Cup. He also researches and writes on typography for the web: he’s author of Responsive Typography from O’Reilly, articles for TYPE Magazine, .Net Magazine, PRINT Magazine, HOW, Monotype.com, and frequent podcast guest. Author of online courses for Aquent’s Gymnasium platform and Frontend Masters. He’s an experienced speaker and workshop leader, having presented at over 50 national and international conferences. The real story: mainly he just follows Tristan and Tillie around Turner Reservoir, posting photos on Instagram.

More articles by Jason

Mandy Michael turns the corner on our variable font adventure and stumbles into a grotto of wonder and amazement. Not forgetting the need for a proper performance budget, Mandy shows how variable fonts can free your creativity from bygone technical constraints.

If you read Jason’s introductory article about variable fonts, you’ll understand the many benefits and opportunities that they offer in modern web development. From this point on we’ll assume that you have either read Jason’s introduction or have some prior knowledge of variable fonts so we can skip over the getting started information. If you haven’t read up on variable fonts before jump over to “Introduction to Variable Fonts: Everything you thought you knew about fonts just changed” first and then come join me back here so we can dive into using variable fonts for interactivity and animations!

Creative Opportunities

If we can use variable fonts to improve the performance of our websites while increasing the amount of style variations available to us, it means that we no longer need to trade off design for performance. Creativity can be the driving force behind our decisions, rather than performance and technical limitations.

My goal is to demonstrate how to create interactive, creative text on the web by combining variable fonts with CSS and JavaScript techniques that you may already be familiar with. With the introduction of variable fonts, designs which would have previously been a heavy burden on performance, or simply impossible due to technical limitations, are now completely possible.

The tone and intent of our words can be more effectively represented with less worry over the impacts of loading in “too many font weights” (or other styles). This means that we can start a new path and focus on representing the content in more meaningful ways. For example, emphasising different words, or phrases depending on their importance in the story or content.

Note: using variable fonts does not negate the need for a good web font performance strategy! This is still important, because after all, they are still fonts. Keep that in mind and check out some of the great work done by Monica Dinculescu, Zach Leatherman or this incredible article by Helen Homes.

Variable Fonts & Animations

Because variable fonts can have an interpolated range of values we can leverage the flexibility and interactive nature of the web. Rather than using SVG, videos or JavaScript to accomplish these effects, we can create animations or transitions using real text, and we can do this using techniques we may already be familiar with. This means we can have editable, selectable, searchable, copy-pastable text, which is accessible via a screenreader.

Grass Variable Font Demo

This effect is achieved using a font called Decovar, by David Berlow. To achieve the animation effect we only need a couple of things to get started.

First, we set up the font-family and make use of the new property font-variation-settings to access the different axes available in Decovar.

h1 {
  font-family: "Decovar";
  font-variation-settings: 'INLN' 1000, 'SWRM' 1000;
}

For this effect, we use two custom axis – the first is called “inline” and is represented by the code INLI and the second is “skeleton worm” represented by the code SWRM. For both axes, the maximum value is 1000 and the minimum value is 0. For this effect, we’ll make the most of the full axis range.

Once we have the base set up, we can create the animation. There are a number of ways to animate variable fonts. In this demo, we’ll use CSS keyframe animations and the font-variation-settings property, but you can also use CSS transitions and JavaScript as well.

The code below will start with the “leaves” expanded and then shrink back until it disappears.

@keyframes grow {
  0% {
    font-variation-settings: 'INLN' 1000, 'SWRM' 1000;
  }
  100% {
    font-variation-settings: 'INLN' 1000, 'SWRM' 0;
  }
}

Once we have created the keyframes we can add the animation to the h1 element, and that is the last piece needed in order to create the animation.

h1 {
  font-family: "Decovar";
  font-variation-settings: 'INLN' 1000, 'SWRM' 1000;
  animation: grow 4s linear alternate infinite;
}

What this demonstrates is that typically, to accomplish effects like this, the heavy lifting is done by the font. We really only need a few lines of CSS for the animation, which if you think about it, is pretty incredible.

There are all sorts of interesting, creative applications of variable fonts, and a lot of incredible fonts you can make the most of. Whether you want to create that “hand-writing” effect that we often see represented with SVG, or something a little different, there are a lot of different options.

Duos Writer: Hand Writing

Decovar: Disappearing Text

Cheee: Snow Text

Variable Fonts, Media Queries and Customisation

It’s not that these are just beautiful or cool effects, what they demonstrate is that as developers and designers we can now control the font itself and that that means is that variable fonts allow typography on the web to adapt to the flexible nature of our screens, environments and devices.

We can even make use of different CSS media queries to provide more control over our designs based on environments, light contrast and colour schemes.

Though the CSS Media Queries Level 5 Spec is still in draft stages, we can experiment with the prefers-color-scheme (also known as dark mode) media query right now!

Dark Mode featuring Oozing Cheee by OhNoTypeCo

The above example uses a font called “Cheee” by OhNoTypeCo and demonstrates how to make use of a CSS Transition and the prefers-color-scheme media query to transition the axis of a variable font.

h1 {
  font-family: “Cheee"
  font-variation-settings: "TEMP" 0;
  transition: all 4s linear;
}

@media (prefers-color-scheme: dark) {
  h1 {
    font-variation-settings: "TEMP" 1000;
  }
}

Dark mode isn’t just about changing the colours, it’s important to consider things like weight as well. It’s the combination of the weight, colour and size of a font that determines how legible and accessible it is for the user. In the example above, I’m creating a fun effect – but more practically, dark mode allows us to modify the contrast and styles to ensure better legibility and usability in different environments.

What is even more exciting about variable fonts in this context is that if developers and designers can have this finer control over our fonts to create more legible, accessible text, it also means the user has access to this as well. As a result, users that create their own stylesheets to customise the experience to their specific requirements, can now adjust the pages font weight, width or other available axis to what best suits them. Providing users with this kind of flexibility is such an incredible opportunity that we have never had before!

As CSS develops, we’ll have access to different environmental and system features that allow us to take advantage of our users unique circumstances. We can start to design our typography to adjust to things like screen width - which might allow us to tweak the font weight, width, optical size or other axes to be more readable on smaller or larger screens. Where the viewport is wide we can have more detail, when its smaller in a more confined space we might look at reducing the width of the font—this helps to maintain the integrity of the design as the viewport gets smaller or, to fit text into a particular space.

We have all been in the situation where we just need the text to be slightly narrower to fit within the available space. If you use a variable font with a width axis you can slightly modify the width to adjust to the space available, and do so in a way that the font was designed to do, rather than using things like letter spacing which doesn’t consider the kerning of the characters.

Variable Fonts, JavaScript and Interactive Effects

We can take these concepts even further and mix in a little JavaScript to make use of a whole suite of different interactions, events, sensors and apis. The best part about this is whether you are using device orientation, light sensors, viewport resizes, scroll events or mouse movement, the base JavaScript doesn’t really change.

To demonstrate this, we’ll use a straightforward example – we’ll match our font weight to the size of our viewport – as the viewport gets smaller, the font weight gets heavier.

We’ll start off by setting our base values. We need to define the minimum and maximum axis values for the font weight, and the minimum and maximum event range, in this case the viewport size. Basically we’re defining the start and end points for both the font and the event.

// Font weight axis range
const minAxisValue = 200
const maxAxisValue = 900

// Viewport range
const minEventValue = 320px
const maxEventValue = 1440px

Next we determine the current viewport width, which we can access with something like window.innerWidth.

// Current viewport width
const windowWidth = window.innerWidth

Using the current viewport width value, we create the new scale for the viewport, so rather than the pixels values we convert it to a range of 0 - 0.99.

const windowSize = (windowWidth - minEventValue) / (maxEventValue - minEventValue)
// Outputs a value from 0 - 0.99

We then take that new viewport decimal value and use it to determine the font weight based on viewport scale.

const fontWeight = windowSize * (minAxisValue - maxAxisValue) + maxAxisValue;
// Outputs a value from 200 - 900 including decimal places

This final value is what we use to update our CSS. You can do this however you want – lately I like to use CSS Custom Properties. This will pass the newly calculated font weight value into our CSS and update the weight as needed.

// JavaScript
p.style.setProperty("--weight", fontWeight);

Finally, we can put all this inside a function and inside an event listener for window resize. You can modify this however you need to in order to improve performance, but in essence, this is all you need to achieve the desired outcome.

function fluidAxisVariation() {
  // Current viewport width
  const windowWidth = window.innerWidth

  // Get new scales for viewport and font weight
  const viewportScale = (windowWidth - 320) / (1440 - 320);
  const fontWeightScale = viewportScale * (200 - 900) + 900;

  // Set in CSS using CSS Custom Property
  p.style.setProperty("--weight", fontWeightScale);
}

window.addEventListener("resize", fluidAxisVariation);

You can apply this to single elements, or multiple. In this case, I’m changing the paragraph font weights and different rates, but also reducing the width axis of the headline so it doesn’t wrap onto multiple lines.

As previously mentioned, this code can be used to create all sorts of really amazing, interesting effects. All that’s required is passing in different event and axis values.

In the following example, I’m using mouse position events to change the direction and rotation of the stretchy slinky effect provided by the font “Whoa” by Scribble Tone.

We can also take the dark mode/colour schemes idea further by making use of the Ambient Light Sensor to modify the font to be more legible and readable in low light environments.

This effect uses Tiny by Jack Halten Fahnestock from Velvetyne Type Foundry and demonstrates how we modify our text based by query the characteristics of the user’s display or light-level, sound or other sensors.

It’s only because Variable fonts give us more control over each of these elements that we can fine-tune the font characteristics to maximise the legibility, readability and overall accessibility of our website text. And while these examples might seem trivial, they are great demonstrations of the possibilities. This is a level of control over our fonts and text that is unprecedented.

Variable Fonts offer a new world of interactivity, usability and accessibility, but they are still a new technology. This means we have the opportunity to figure out how and what we can achieve with them. From where I stand, the possibilities are endless, so don’t be limited by what we can already do – the web is still young and there is so much for us to create. Variable fonts open up doors that never existed before and they give us an opportunity to think more creatively about how we can create better experiences for our users.

At the very least, we can improve the performance of our websites, but at best, we can make more usable, more accessible, and more meaningful content - and that, is what gets me really excited about the future of web typography with variable fonts.

About the author

Mandy is a community organiser, speaker, and developer working as the Front End Development Manager at Seven West Media in Western Australia. She is a co-organiser and Director of Mixin Conf, and the founder and co-organiser of Fenders, a local meetup for front-end developers providing events, mentoring and support to the Perth web community.

Mandy’s passion is CSS, HTML and JS and hopes to inspire that passion in others. She loves the supportive and collaborative nature of the web and strives to encourage this environment through the community groups she is a part of. Her aim is to create a community of web developers who can share, mentor, learn and grow together.

More articles by Mandy

Frances Berriman asks us to give the gift of consideration to those who are using the web on constricted devices such as low-end smart phones or feature phones. Christmas is a time of good will to all, and as Bugsy Malone reminds us, you give a little love and it all comes back to you.

If I was given the job of Father Christmas with all my human limitations, apparently it would take me something like 6 months at non-stop full speed to deliver gifts to every kid on the planet. The real Father Christmas has the luxury of magic when it comes to delivering millions of gifts in just one night, but the only magical platform at my disposal is the world wide web, so I propose switching to digital gift cards and saving the reindeer feed.

300 million people are set to come online for the very first time in 2020, and a majority of those will be doing so via mobile phones (smart- and feature-phones). If we want those new users to have a great time online, spending those gift cards, we need to start thinking about their needs and limitations.

Suit up

We might not be hopping on the sleigh for these deliveries, but let’s suit up for the journey and get the tools we need to start testing and checking how our online gift-receivers will be enjoying their online shopping experience.

Of course, the variety of phones and OSs out there is huge and varied, but we have a few options out there to get a sense for the median. Here’s a few suggestions on where to start:

• Never has there been a better time to advocate at your workplace for a device testing suite or lab.
• You can also just pick up a low-end phone for a few bucks and spend some real time using it and getting a sense for how it feels to live with it every day. May I suggest the Nokia 2 or the Moto E6 - both very representative devices of the sort our new visitors will be on.
• You’ve also got WebPageTest.org at your disposal, where you can emulate various phones and see your sites rendered in real-time to get a sense of what an experience may look like for your users.
• You’ll also want to set yourself some goals. A performance budget, for example, is a good way to know if the code you’re shipping hits the mark in a more programmatic way.

Gift wrap

Many of us began our internet lives on desktop machines, and thanks to Moore’s law, these machines have been getting ever more powerful every year with more CPUs and memory at our disposal. The mobile phone landscape somewhat resets us on what hardware capacity is available on the client-side of our code, so it’s time to lighten the load.

What we see in the landscape of phones today is a huge spread of capabilities and CPU speeds, storage capacity and memory. And the gap between the haves and the have-nots is widening, so we have a huge task to deal with in meeting the needs of such a varied audience.

As far as possible, we should try to:

• Keep processing off the client - do anything you can server-side. Consider a server-side render (hold the <script>, thanks) for anything relatively static (including cached frequent queries and results) to keep client-side JavaScript to the minimum. This way you’re spending your CPU, not the user’s.
• Avoid sending everything you have to to the end user. Mobile-first access also means data-plan-first access for many, which means they may be literally paying in cold-hard cash for everything you send over the wire – or may be experiencing your site over a degraded “4G” connection towards the end of the month.
• Aggressively cache assets to prevent re-downloading anything you’ve sent before. Don’t make the user pay twice if they don’t have to.
• Progressively load additional assets and information as the user requests them, rather than a big upfront payload, that way you’re giving the end user a little more choice about whether they want or need that extra data set.

This is all to say that as web developers, we have a lot more control over how and when we deliver the meat of our products - unlike native apps that generally send the whole experience down as one multi-megabyte download that our 4G and data-strapped users can’t afford.

Make a wish

Finally, it’s time for your gift recipients to go out onto the web and find whatever their greatest wish is. For many, that’s going to begin when they first turn on their phone and see all those enticing icons on their home screen. Opening a browser may not be their first port of call.

They’ll be primed to look for sites and information through the icon-heavy menu that most mobile OSs use today, and they will be encouraged to find new experiences through the provided app store interface.

The good news is that web experience can be found in many modern app-stores today.

For example, if you build an app using Trusted Web Activities, the Google Play Store will list your web site right alongside native apps and allow users to install them on their phones. Samsung and Microsoft have similar options without the extra step of creating a TWA - they’ll list any Progressive Web App in their stores. Tools like Microsoft’s PWA Builder and Llama Pack are making this easier than ever.

If your users are primed to search for new experiences via a search engine instead, then they’ll benefit from the work you’ve put in to list them in app stores regardless, as PWAs are first and foremost about making websites mobile-friendly, regardless of point of sale. A PWA will provide them with offline support, service works, notifications and much more.

We do have a grinch in this story, however.

Apple’s iOS explicitly does not allow your website to be listed in their app store, so sadly you’ll have a harder time reaching those users. But it is possible! Fortunately, iOS isn’t as all-dominating world wide as it is in the tech community, selling only around 10-15% of smartphones out in the world.

The best present

The WWW is a wonderful gift that we received over 30 years ago and, as web developers, we get to steward and share this truly global, open, platform with millions of people every day. Let’s take care of it by building and sharing experiences that truly meet the needs of everyone.

About the author

Frances Berriman is a San Francisco-based British-born designer and web developer who blogs at fberriman.com. She’s done all sorts of things, but has a special soft spot for public sector projects, and has worked for the Government Digital Service, building GOV.UK, Code for America, Nature Publishing and the BBC and is currently Head of UX and Product Design at Netlify.

More articles by Frances

Colin Bendell gets into the minutia of microbrowsers - the small previews of your site that are pervasive all around the web and through social media apps and search engines whenever an item of content on your site is referenced.

You’ve seen it everywhere - that little thumbnail preview of a website mentioned in a tweet, the expanded description in a Slack channel, or in WhatsApp group chat.

Figure 1: The preview shown in a group chat provides a hint of what the real webpage looks like

These link previews are so commonplace that we hardly pay any attention to how our site design might be impacting the generated preview. Yet, these previews can be the most influential part for attracting new audiences and increasing engagement - possibly more than SEO. Even more alarming is that most web analytics are blind to this traffic and can’t show you how these Microbrowsers are interacting with your site.

As we close out the year, here are five essential questions and ideas that every web dev should know about Microbrowsers.

1. What are Microbrowsers? How are they different from “normal” browser?

We are all very familiar with the main browsers like Firefox, Safari, Chrome, Edge and Internet Explorer. Not to mention the many new browsers that use Chromium as the rendering engine but offer unique user experiences like Samsung Internet or Brave.

In contrast, Microbrowsers are a class of User-Agents that also visit website links, parse HTML and generate a user experience. But unlike those traditional browsers, the HTML parsing is limited and the rendering engine is singularly focused. The experience is not intended to be interactive. Rather the experience is intended to be representational - to give the user a hint of what exists on the other side of the URL.

Creating link previews is not new. Facebook and Twitter have been adding these link previews in posts for nearly a decade. That used to be the primary use case. Marketing teams created backlog items to adopt different microdata - from Twitter Cards and Open Graph annotations for Facebook. LinkedIn likewise embraced both Open Graph and OEmbed tags to help generate the previews

<meta name="description" content="seo description long">
<meta name="keywords" content="seo keyword list">

<link rel="shortcut icon" href="favicon.ico" 
                          type="image/x-icon">
<link rel="icon" href="favicon_32.png" sizes="32x32">
<link rel="icon" href="favicon_48.png" sizes="48x48">
<link rel="icon" href="favicon_96.png" sizes="96x96">
<link rel="icon" href="favicon_144.png" sizes="144x144">

<meta property="og:title" content="Short title here" />
<meta property="og:description" content="shortish description" />
<meta name="twitter:title" content="Short title here">
<meta name="twitter:description" content="shortish description">

<meta property="og:image"
      content="https://res.cloudinary.com/.../hero-img.png" />

<meta name="twitter:image:src"
      content="https://res.cloudinary.com/.../hero-img.png">

As group chats and other collaboration tools have become more prevalent, we have seen many features from the big social media platforms emerge. Particularly in recent years we’ve seen the adoption of the link unfurling behaviour in these chat platforms. Rather than reinventing the wheel, each platform looks for pre-existing microdata to generate the preview.

But which data should be used? How should this be arranged? As it turns out, each platform behaves slightly differently; presenting information in slightly different ways.

Figure 2: The same amazon link shared in iMessage (left), Hangouts and WhatsApp (right)

2. If Microbrowsers are everywhere, why don’t I see them in my analytics reports?

It’s easy to miss the traffic from Microbrowsers. This is for a number of reasons:

First, page requests from Microbrowsers don’t run JavaScript and they don’t accept cookies. The Google Analytics <script> block won’t be run or executed. And all cookie will be ignored by the rendering agent.

Second, if you were to do a log analysis based on HTTP logs from your CDN or web stack, you would see a relatively small volume of traffic. That is assuming you can identify the User-Agent strings. Some of these Microbrowsers impersonate real browsers and others impersonate Facebook or twitter. For example, iMessage uses the same User-Agent string for all these requests and it hasn’t changed since iOS 9.

User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_11_1) 
             AppleWebKit/601.2.4 (KHTML, like Gecko) 
             Version/9.0.1 Safari/601.2.4 
             facebookexternalhit/1.1  
             Facebot Twitterbot/1.0

Finally, many platforms - particularly Facebook Messenger and Hangouts use centralized services to request the preview layout. This, in contrast to WhatsApp and iMessage where you will see one request per user. In the centralized consumer approach your web servers will only see one request, but this one request might represent thousands of eyeballs.

3. Microbrowser are probably more important than google bot

We all know the importance of having our web sites crawled by search engines like googlebot. These bots are the lifeblood for lead generation and for discovering new users.

However, the real gold for marketers is from word-of-mouth discussions. Those conversations with your friends when you recommend a TV show, a brand of clothing, or share a news report. This is the most valuable kind of marketing.

Last year when assembling the data for Cloudinary’s State of the Visual Media report, I discovered that there was a very prominent usage pattern over the USA holiday season. During thanksgiving, all the way to Black Friday, the rate of link sharing skyrocketed as group chats shared deals and insights.

Zooming out (and normalizing for time-of-day), we can see that there is a daily cadence of link sharing and word of mouth referrals. It probably isn’t a shock to see that we predominantly share links in Slack between Monday and Friday, while WhatsApp is used all week long. Likewise, WhatsApp is most often used during our ‘break’ times like lunch or in the evening after we put the kids to bed.

While the link preview is increasingly common, there are two user behaviours to balance:

• Users can be skeptical of links sent via SMS and other chats. We don’t want to be fooled into clicking a phishing links and so we look for other queues to offer validation. This is why most platforms use the preview while also emphasize the website url host name.

• Skimming. I’m sure you’ve had the experience coming out of a meeting or grocery store to find a group chat with 100 messages. As you scroll to catch up on the conversation, links can easily be skipped. In this way, users expect the preview to act as a summary to tell them how important it is to visit the link.

Figure 4: Nielsen Norman Group summarizes the research in a dynamic image preview

Figure 5: A mockup of how an ecommerce product could create compelling previews showcasing colors, stock and price in the preview

4. Microbrowsers are not real browsers (they just play one on TV)

As I previously mentioned, Microbrowsers pretend to be a browser in that they send the right HTTP headers and often send impersonating User-Agent strings. Yet, there are several characteristics that a web dev should be aware of.

First, Microbrowsers try to protect the User’s privacy. The user hasn’t decided to visit your site yet, and more importantly, the user is having a private conversation. The fact that your brand or website is mentioned should just make your ears burn, but you shouldn’t be able to listen in to the conversation.

For this reason, all Microbrowsers:

• don’t execute JavaScript - so your react application won’t work
• ignore all cookies - so your A/B or red/green cookies will be ignored
• some will follow redirects, but will quickly time out after a few seconds and give up trying to expand the link.
• there won’t be a referer: HTTP header when the user clicks the link for the full browser. In fact, a new user will appear as ‘direct’ traffic - as though they typed in the url.

Second, Microbrowsers have a very small brain and very likely don’t use an advanced network algorithm. Most browsers will use a tokenizer to parse the HTML markup and send requests to the network stack asynchronously. Better yet, browsers will do some analysis of the resources needed before sending the async request to the network.

Based on observational experimentation, most platforms simply use a glorified for loop when parsing the HTML and often request the resources synchronously. This might be ok for fast wifi experiences, but it can cause inconsistent experiences on flaky wifi.

For example, iMessage will discover and load all <link rel="icon" > favicon, all <meta property="og:image" images, and all <meta name="twitter:image:src" before deciding what to render. Many sites still advertise 5 or more favicon sizes. This means that iMessage will download all favicons regardless of size and then not use them if it decides to instead render the image.

For this reason the meta markup that is included is important. The lighter the content, the more likely it will be to be rendered.

5. Markup Matters

Since Microbrowsers are simple-brained browsers, it is all the more important to produce good markup. Here are a few good strategies:

• It’s almost 2020, you only need one favicon size. Remove all the other <link rel="shortcut icon" and <link rel="icon" references.
• Based on observational experimentation, the most commonly recognized microdata tags for preview are the Open-Graph tags. When the OG and twitter card tags are missing, the default SEO <meta name="description" is used. However, since the description is often nonsensical SEO optimized phrases, users’ eyes will likely glaze over.

• On that note, use good descriptive text
• Provide up to three <meta property="og:image" images. Most platforms will only load the first one, while others (notably iMessage) attempts to create a collage.

Figure 6: Amazon uses User-Agent detection which results in many link previews using the description meta tag.

• Use <meta property="og:video* with progressive (not streaming) video experiences.

<meta property="og:type" content="video.other">
<meta property="og:video:url" 
               content="https://shoesbycolin.com/blue.mp4">
<meta property="og:video:secure_url" 
               content="https://shoesbycolin.com/blue.mp4">
<meta property="og:video:type" content="video/mp4">
<meta property="og:video:width" content="1280">
<meta property="og:video:height" content="720">

• Don’t use UA sniffing to hide the <meta> tags. Sites like Amazon do this to try and show only Facebook/Twitter the microdata annotated website. But this can cause problems for some Microbrowsers that don’t use the same impersonation convention. The result is a simple link without a preview.
• Use the opportunity to tell your product story or summarize your ideas.

Summary

As more of our conversations happen in group chats and slack channels, link previews are an important way for you to engage users before they start the journey on your site. Unfortunately, not all websites present good or compelling previews. (And now that you know what to look for, you won’t be able to unsee bad examples - I’m sorry). To help users take the leap and visit your site, we need to make sure that all our pages are annotated with microdata. Better yet, we can use these previews to create compelling visual summaries.

About the author

Colin is part of the CTO Office at Cloudinary and co-author of the O’Reilly book High Performance Images. He spends much of his time at the intersection of high volume data, media, browsers and standards. He recently helped the community effort writing chapters in the Web Almanac on Media and CDNs. You can find him on tweeting @colinbendell and at blogging at https://bendell.ca

More articles by Colin

Chris Ferdinandi turns the heat down low and lets the sauce reduce while we take a look at how to add spice to our source with a sprinkling of Array.reduce(). Just a little ingenuity with the humblest of functions.

Of all the modern array methods, the one I had the hardest time wrapping my head around was Array.reduce().

On the surface, it seems like a simple, boring method that doesn’t do much. But below its humble exterior, Array.reduce() is actually a powerful, flexible addition to your developer toolkit.

Today, we’re going to look at some cool things you can do with Array.reduce().

How Array.reduce() works

Most of the modern array methods return a new array. The Array.reduce() method is a bit more flexible. It can return anything. Its purpose is to take an array and condense its content into a single value.

That value can be a number, a string, or even an object or new array. That’s the part that’s always tripped me up – I didn’t realize just how flexible it is!

The syntax

The Array.reduce() accepts two arguments: a callback method to run against each item in the array, and a starting value.

The callback also accepts two arguments: the accumulator, which is the current combined value, and the current item in the loop. Whatever you return is used as the accumulator for the next item in the loop. On the very first loop, that starting value is used instead.

var myNewArray = [].reduce(function (accumulator, current) {
  return accumulator;
}, starting);

Let’s look at some examples to make this all tangible.

1. Adding numbers together

Let’s say you had an array of numbers that you wanted to add together. Using Array.forEach(), you might do something like this:

var total = 0;

[1, 2, 3].forEach(function (num) {
  total += num;
});

This is the cliche example for using Array.reduce(). I find the word accumulator confusing, so in this example, I’m calling it sum, because that’s what it is.

var total = [1, 2, 3].reduce(function (sum, current) {
  return sum + current;
}, 0);

Here, we pass in 0 as our starting value.

In the callback, we add the current value to the sum, which has our starting value of 0 on the first loop, then 1 (the starting value of 0 plus the item value of 1), then 3 (the sum value of 1 plus the item value of 2), and so on.

Here’s a demo.

2. Combining multiple array methods into Array.map() and Array.filter() into a single step

Imagine you had an array of wizards at Hogwarts.

var wizards = [
  {
    name: 'Harry Potter',
    house: 'Gryfindor'
  },
  {
    name: 'Cedric Diggory',
    house: 'Hufflepuff'
  },
  {
    name: 'Tonks',
    house: 'Hufflepuff'
  },
  {
    name: 'Ronald Weasley',
    house: 'Gryfindor'
  },
  {
    name: 'Hermione Granger',
    house: 'Gryfindor'
  }
];

You want to create a new array that contains just the names of wizards who are in Hufflepuff. One way you could do that is by using the Array.filter() method to get back just wizards whose house property is Hufflepuff. Then, you’d use the Array.map() method to create a new array containing just the name property for the remaining wizards.

// Get the names of the wizards in Hufflepuff
var hufflepuff = wizards.filter(function (wizard) {
  return wizard.house === 'Hufflepuff';
}).map(function (wizard) {
  return wizard.name;
});

With the Array.reduce() method, we can get the same array in a single pass, improving our performance. You pass in an empty array ([]) as the starting value. On each pass, you check to see if the wizard.house is Hufflepuff. If it is, you push it to the newArr (our accumulator in this example). If not, you do nothing.

Either way, you return the newArr to become the accumulator on the next pass.

// Get the names of the wizards in Hufflepuff
var hufflepuff = wizards.reduce(function (newArr, wizard) {
  if (wizard.house === 'Hufflepuff') {
    newArr.push(wizard.name);
  }
  return newArr;
}, []);

Here’s another demo.

3. Creating markup from an array

What if, instead of creating an array of names, we wanted to create an unordered list of wizards in Hufflepuff? Instead of passing an empty array into Array.reduce() as our starting value, we’ll pass in an empty string ('') and call it html.

If the wizard.house equals Hufflepuff, we’ll concatenate our html string with the wizard.name wrapped in an opening and closing list item (li). Then, we’ll return the html to become the accumulator on the next loop.

// Create a list of wizards in Hufflepuff
var hufflepuffList = wizards.reduce(function (html, wizard) {
  if (wizard.house === 'Hufflepuff') {
    html += '<li>' + wizard.name + '</li>';
  }
  return html;
}, '');

Add an opening and closing unordered list element before and after Array.reduce(), and you’re ready to inject your markup string into the DOM.

// Create a list of wizards in Hufflepuff
var hufflepuffList = '<ul>' + wizards.reduce(function (html, wizard) {
  if (wizard.house === 'Hufflepuff') {
    html += '<li>' + wizard.name + '</li>';
  }
  return html;
}, '') + '</ul>';

See it in action here.

4. Grouping similar items in an array together

The lodash library has a groupBy() method takes a collection of items as an array and groups them together into an object based on some criteria.

Let’s say you want an array of numbers.

If you wanted to group all of the items in numbers together based on their integer value, you would do this with lodash.

var numbers = [6.1, 4.2, 6.3];

// returns {'4': [4.2], '6': [6.1, 6.3]}
_.groupBy(numbers, Math.floor);

If you had an array of words, and you wanted to group the items in words by their length, you would do this.

var words = ['one', 'two', 'three'];

// returns {'3': ['one', 'two'], '5': ['three']}
_.groupBy(words, 'length');

Creating a groupBy() function with Array.reduce()

You can recreate that same functionality using the Array.reduce() method.

We’ll create a helper function, groupBy(), that accepts the array and criteria to sort by as arguments. Inside groupBy(), we’ll run Array.reduce() on our array, passing in an empty object ({}) as our starting point, and return the result.

var groupBy = function (arr, criteria) {
  return arr.reduce(function (obj, item) {
    // Some code will go here...
  }, {});
};

Inside the Array.reduce() callback function, we’ll check to see if the criteria is a function, or a property of the item. Then we’ll get its value from the current item.

If there’s no property in the obj with that value yet, we’ll create it and assign an empty array as its value. Finally, we’ll push the item to that key, and return the object as the accumulator for the next loop.

var groupBy = function (arr, criteria) {
  return arr.reduce(function (obj, item) {

    // Check if the criteria is a function to run on the item or a property of it
    var key = typeof criteria === 'function' ? criteria(item) : item[criteria];

    // If the key doesn't exist yet, create it
    if (!obj.hasOwnProperty(key)) {
      obj[key] = [];
    }

    // Push the value to the object
    obj[key].push(item);

    // Return the object to the next item in the loop
    return obj;

  }, {});
};

Here’s a demo of the completed helper function.

Special thanks to Tom Bremer for helping me make some improvements to this one. You can find this helper function and more like it on the Vanilla JS Toolkit.

5. Combining data from two sources into an array

Remember our array of wizards?

var wizards = [
  {
    name: 'Harry Potter',
    house: 'Gryfindor'
  },
  {
    name: 'Cedric Diggory',
    house: 'Hufflepuff'
  },
  {
    name: 'Tonks',
    house: 'Hufflepuff'
  },
  {
    name: 'Ronald Weasley',
    house: 'Gryfindor'
  },
  {
    name: 'Hermione Granger',
    house: 'Gryfindor'
  }
];

What if you had another data set, an object of house points each wizard has earned.

var points = {
  HarryPotter: 500,
  CedricDiggory: 750,
  RonaldWeasley: 100,
  HermioneGranger: 1270
};

Imagine you wanted to combine both sets of data into a single array, with the number of points added to each wizard’s data in the wizards array. How would you do it?

The Array.reduce() method is perfect for this!

var wizardsWithPoints = wizards.reduce(function (arr, wizard) {

  // Get the key for the points object by removing spaces from the wizard's name
  var key = wizard.name.replace(' ', '');

  // If the wizard has points, add them
  // Otherwise, set them to 0
  if (points[key]) {
    wizard.points = points[key];
  } else {
    wizard.points = 0;
  }

  // Push the wizard object to the new array
  arr.push(wizard);

  // Return the array
  return arr;

}, []);

Here’s a demo combining data from two sources into an array.

6. Combining data from two sources into an object

What if you instead wanted to combine the two data sources into an object, where each wizard’s name was the key, and their house and points were properties? Again, the Array.reduce() method is perfect for this.

var wizardsAsAnObject = wizards.reduce(function (obj, wizard) {

  // Get the key for the points object by removing spaces from the wizard's name
  var key = wizard.name.replace(' ', '');

  // If the wizard has points, add them
  // Otherwise, set them to 0
  if (points[key]) {
    wizard.points = points[key];
  } else {
    wizard.points = 0;
  }

  // Remove the name property
  delete wizard.name;

  // Add wizard data to the new object
  obj[key] = wizard;

  // Return the array
  return obj;

}, {});

Here’s a demo combining two data sets into an object.

Should you use Array.reduce() more?

The Array.reduce() method has gone from being something I thought was pointless to my favorite JavaScript method. So, should you use it? And when?

The Array.reduce() method has fantastic browser support. It works in all modern browsers, and IE9 and above. It’s been supported in mobile browsers for a long time, too. If you need to go back even further than that, you can add a polyfill to push support back to IE6.

The biggest complaint you can make about Array.reduce() is that it’s confusing for people who have never encountered it before. Combining Array.filter() with Array.map() is slower to run and involves extra steps, but it’s easier to read. It’s obvious from the names of the methods what they’re supposed to be doing.

That said, there are times where Array.reduce() makes things that would be complicated more simple rather than more complicated. The groupBy() helper function is a good example.

Ultimately, this is another tool to add to your toolkit. A tool that, if used right, can give you super powers.

About the author

Chris Ferdinandi helps people learn vanilla JavaScript. He believes there’s a simpler, more resilient way to make things for the web.

Chris is the author of the Vanilla JS Pocket Guide series, creator of the Vanilla JS Academy training program, and host of the Vanilla JS Podcast. His developer tips newsletter is read by thousands of developers each weekday.

He’s taught developers at organizations like Chobani and the Boston Globe, and his JavaScript plugins have been used used by Apple and Harvard Business School. Chris Coyier, the founder of CSS-Tricks and CodePen, has described his writing as “infinitely quote-worthy.”

Chris loves pirates, puppies, and Pixar movies, and lives near horse farms in rural Massachusetts. He runs Go Make Things with Bailey Puppy, a lab-mix from Tennessee.

More articles by Chris

Cambridge, Mass. : National Bureau of Economic Research, 2020

Hoboken : Taylor & Francis, 2003.

London : Pluto Press, 2020.

Cambridge, Mass. : National Bureau of Economic Research, 2020

Dewey Library - HV7561.H85 2020

Dewey Library - U39.C76 2020

Dewey Library - HV9506.R53 2019

Online Resource

Dewey Library - JK2356.G76 2019

Dewey Library - HV6692.T39 L48 2019

Dewey Library - JK2071.H57 2019

Dewey Library - JC571.C78 2019

Dewey Library - JC573.2.U6 W54 2019

Dewey Library - JC573.2.U6 D8354 2020

Dewey Library - HV6626.2.S59 2019

Online Resource

Online Resource

Online Resource

Dewey Library - JC491.K255 2020

Dewey Library - HV6432.B364 2019