Browsers - Safari and WebKit

What is the Value of Browser Diversity?

In May 2019 I attended a talk by Mike Taylor who works on webcompat at Mozilla. Mike told the sordid story of window.event, a non-standard IE invention that was replicated in Konqueror, which showed up in Webkit, which stuck around in Blink, and was now Mike’s problem in Firefox. It was a good story fraught with ups and downs and literal “Breaking the Web” level changes for a tiny feature rollout.

At the end of the talk Mike threw out a pretty prescient question (Edge had just released its Chromium beta) and I’ve been mulling over it ever since:

What is the value of browser diversity? If Firefox switched to Chromium tomorrow, what would we lose?

– Mike Taylor, a traitor (apparently)

Mike made it clear that the question was purely theoretical and no serious talk about this was happening at Mozilla at the time. Nonetheless, it was a challenging thought. Throwing away all sunk costs, what is the value of the colossal expense required to employ engineering teams to chase Chrome’s tail?

I’ve thought about these questions for over a year and narrowed my feelings of browser diversity down to two major value propositions:

  1. Browser diversity keeps the Web deliberately slow
  2. Browser diversity fosters consensus and cooperation over corporate rule

They are similar, but slightly different concepts for me.

Slow, like brisket.

I think the Web platform’s most frustrating aspect is also its greatest asset: it’s slow. It’s not just slow, it’s “it took 10 years to ship the <main> element which is just a spicy <div>” kind of slow. It’s glacial.

This can be agonizing while you wait for a much needed feature to roll out in all browsers, only to find out five years in the process one browser refuses the entire premise of the feature (RIP HTML Imports). The big tradeoff is that web platform features have to run the gauntlet and more thinking is applied over time: robustness, naming, internationalization, accessibility, security, etc. all have proper time for consideration and aren’t rushed through like it’s a product sprint.

[…]

Cooperation, not corporation.

Browser diversity means browser vendors are hindered from shipping features that only benefit themselves (e.g. ActiveX in IE, -webkit-box-reflect, etc). Good ideas have to be open and meet a pseudo-requirement of ubiquitous utility. To make good platform features requires consensus, not competition, and the web’s potential is born out of cooperation, not a single corporation.

It’s hard to quantify this, but if all aspects of the Web (development, browsing, searching, hosting) are ceded to a single corporation, all it takes is one heavy-handed manager or executive hellbent on hitting some OKRs to push their thumb on the scale of the Web. If the Web is governed by a single corporation, it will start looking like that corporation’s vision of the Web, ultimately limiting its own potential. Trading short term gain on new shiny features for long term vision.

Why browser diversity matters: Chrome unilaterally creates de facto standards

Yet another great example of why browser diversity matters and why Chrome’s overwhelming presence in both mobile and desktop use is harmful to the open web: some developers mistake Chrome’s adoption of an API as a web standard, when both Mozilla and Apple have serious concerns about the security of said API:

In issue #509 of JavaScript Weekly, Chrome’s new File System Access API was mistakenly referred to as an “open standard.” The author probably assumed that a feature with a specification and an implementation in Chrome must therefore be a web standard, but that is not necessarily the case.

The API in question is currently hosted by the Web Incubator Community Group (WICG), a place where browser vendors can propose, discuss, and develop new web platform features, and receive feedback from the wider community.

[…]

Google has been developing the File System Access API for at least the past two years and decided to ship it in Chrome in October (last month). As part of this process, Google asked both Apple and Mozilla for their official positions on the API. So far, their responses have not been positive (Apple, Mozilla).

It seems that Google decided to ship the File System Access API in Chrome without endorsement from Apple or Mozilla because it believes that this feature “moves the web platform forward”:

Interoperability risk is the risk that browsers will not eventually converge on an interoperable implementation of the proposed feature. … If a change has high interop/compat risk but is expected to significantly move the web forward, Chromium will sometimes welcome it.

Standardization and support from Apple or Mozilla is not a requirement for shipping a web platform feature in Chrome. However, because of Chrome’s large market share, there is a risk of such a feature becoming a de facto standard:

Changes to Chrome’s functionality create de facto standards. Market participants must adhere to these standards or risk their technology no longer being compatible with most websites.

Apple's subtle anti-competitive practices and the web

Apple has a history of stunting the web’s progress on its platforms. On iOS, Apple doesn’t allow fully independent third-party browsers, requiring all apps to leverage its Safari browser when rendering web-based content. While browsers like Chrome and Opera are available in the App Store, they must use Apple’s Safari browser behind the scenes to render web pages, rather than their own. That means Apple has a monopoly on how iPhone and iPad users access the web. To push developers toward building native apps on iOS rather than using web technologies, Apple ignores popular parts of the open web specification that other browsers implement, to its own benefit.

A technology called WebRTC, for example, allows video calling in a web browser without additional software. It powers tools like Google Meet. But Apple was incredibly slow to implement the specification, leaving out key pieces of functionality, and the technology didn’t work when embedded inside apps.

Apple also handicapped an emerging standard called Progressive Web Apps (PWAs) — which, like Electron, allows developers to build native-like apps for both desktop and mobile — by partially implementing it in a way that makes it too inconsistent to rely on. PWA doesn’t have the same problem if users open apps in Chrome or Firefox, but iPhone and iPad users can’t install third-party browsers, which makes PWA-based technology a non-starter.

[…]

Apple’s subtle, anti-competitive practices don’t look terrible in isolation, but together they form a clear strategy: Make it so painful to build with web-based technology on Apple platforms that developers won’t bother. Now that the App Store is not accepting apps built using Electron, developers will likely find creative ways to work around it, but Apple is setting up for a continual cat-and-mouse game as it plans to exert more control over which apps can run on the platform in the future.

Mike Taylor - A case for browser-engine diversity... - View Source 2019

Given the recent demise of Presto and EdgeHTML rendering engines, and dominant market share growth for Chrome (and its Chromium engine), can we make a case for browser engine diversity in a decreasingly diverse browser engine world? In this talk, we’ll talk about web compatibility, interoperability, the web standards process, and hopefully conclude that we should care about these things in 2019.

Cutting the mustard with only CSS

JavaScript can be pretty brittle, so having a way to exclude browsers that don’t cut the mustard via CSS can be really useful, especially if you don’t want to serve them large amounts of CSS that they won’t properly understand. Since we can’t prevent loading a stylesheet via feature queries, the media attribute on a <link> element seems the next best thing. Andy Kirk has come up with a few combinations:

Code language: HTML

<!--
  Print (Edge doesn't apply to print otherwise)
  IE 10, 11
  Edge
  Chrome 29+, Opera 16+, Safari 6.1+, iOS 7+, Android ~4.4+
  FF 29+
-->
<link rel="stylesheet" href="your-stylesheet.css" media="
  only print,
  only all and (-ms-high-contrast: none), only all and (-ms-high-contrast: active),
  only all and (pointer: fine), only all and (pointer: coarse), only all and (pointer: none),
  only all and (-webkit-min-device-pixel-ratio:0) and (min-color-index:0),
  only all and (min--moz-device-pixel-ratio:0) and (min-resolution: 3e1dpcm)
">
 
<!--
  Print (Edge doesn't apply to print otherwise)
  Edge, Chrome 39+, Opera 26+, Safari 9+, iOS 9+, Android ~5+, Android UCBrowser ~11.8+
  FF 47+
-->
<link rel="stylesheet" href="your-stylesheet.css" media="
  only print,
  only all and (pointer: fine), only all and (pointer: coarse), only all and (pointer: none),
  only all and (min--moz-device-pixel-ratio:0) and (display-mode:browser), (min--moz-device-pixel-ratio:0) and (display-mode:fullscreen)
">

Safari bug with gradients that fade to "transparent"

Say you have a gradient in CSS that goes from red to transparent. Easy, right? Like this:

Code language: CSS

.element {
  background: linear-gradient(
    to bottom,
    red, 
    transparent
  );
}

There is a pretty big gotcha here, though.

In Chrome (also Android), Firefox, and Edge, you’d be all good.

A side by side comparison of Chrome, Edge, and Firefox, all displaying the same test page containing two rectangles, which have a gradient background that starts red at the top and fades to transparent at the bottom of each rectangle.

But in Safari (also iOS), you’d not be good.

The same test page as before, viewed in both the iOS simulator version of Safari, and the macOS version of Safari. The two rectangles, each with a gradient background colour are different this time. The red in the rectangle on the left fades from red to black, as it fades to transparent, which appears to be a bug. The rectangle on the right fades to transparent while still maintaining the red colour, as expected.
The element on the left in each browser demonstrates the problem.

The problem, the best I understand it, is that transparent is being interpreted (and interpolated) as “transparent black”.

To fix it, you have to set the color as a fully transparent version of that exact color. Like:

Code language: CSS

.element {
  background: linear-gradient(
    to bottom,
    red,
    rgba(255, 0, 0, 0)
  )
}

[…]

Sass can help out, if you are using that:

Code language: Sass

.element {
  background: linear-gradient(
    to bottom,
    #eb8fa9,
    rgba(#eb8fa9, 0);
  )
}