Build a cross-browser extension
The introduction of the browser extensions API created a uniform landscape for the development of browser extensions. However, there are differences in the API implementations and the scope of coverage among the browsers that use the extensions API (the major ones being Chrome, Edge, Firefox, Opera, and Safari).
Maximizing the reach of your browser extension means developing it for at least two browsers, possibly more. This article looks at the main challenges faced when creating a cross-browser extension and suggests how to address these challenges.
Note: The main browsers have adopted Manifest V3. This manifest version provides better compatibility between the browser extension environments, such as promises for handling asynchronous events. In addition to the information in this guide, refer to the Manifest V3 migration guides for Firefox and Chrome.
Cross-platform extension coding hurdles
You need to address the following areas when tackling a cross-platform extension:
API namespace
There are two API namespaces in use among the main browsers:
browser.*
, the proposed standard for the extensions API used by Firefox and Safari.chrome.*
used by Chrome, Opera, and Edge.
Firefox also supports the chrome.*
namespace for APIs that are compatible with Chrome, primarily to assist with porting. However, using the browser.*
namespace is preferred. In addition to being the proposed standard, browser.*
uses promises—a modern and convenient mechanism for handling asynchronous events.
Only in the most trivial extensions is namespace likely to be the only cross-platform issue to be addressed. Therefore, it's rarely, if ever, helpful to address this issue alone. The best approach is to address this with asynchronous event handling.
API asynchronous event handling
With the introduction of Manifest V3, all the main browsers adopted the standard of returning Promises from asynchronous methods. Firefox and Safari have full support for Promises on all asynchronous APIs. Starting from Chrome 121, all asynchronous extension APIs support promises unless documented otherwise. The devtools
API is the only API namespace without Promise support (Chromium bug 1510416).
In Manifest V2, Firefox and Safari support Promises for asynchronous methods. At the same time, Chrome methods invoke callbacks. For compatibility, all the main browsers support callbacks across all manifest versions. See Callbacks and Promises for details.
Some handlers of extension API events are expected to respond asynchronously through a Promise
or callback function. For example, a handler of the runtime.onMessage
event can send an asynchronous response using a Promise
or using a callback. A Promise
as the return value from an event handler is supported in Firefox and Safari, but not yet in Chrome.
Firefox also supports callbacks for the APIs that support the chrome.*
namespace. However, using promises is recommended. Promises greatly simplifies asynchronous event handling, particularly where you need to chain events together. This means using a polyfill or similar so your extension uses the browser.*
namespace in Firefox and Safari and chrome.*
in Chrome, Opera, and Edge.
Note: If you're unfamiliar with the differences between these two methods, look at Getting to know asynchronous JavaScript: Callbacks, Promises and Async/Await or the MDN Using promises page.
The WebExtension browser API Polyfill
So, how do you take advantage of promises easily? The solution is to code for Firefox using promises and use the WebExtension browser API Polyfill to address Chrome, Opera, and Edge.
This polyfill addresses the API namespace and asynchronous event handling across Firefox, Chrome, Opera, and Edge.
To use the polyfill, install it into your development environment using npm or download it directly from GitHub releases.
Then, reference browser-polyfill.js
in:
manifest.json
to make it available to background and content scripts.- HTML documents, such as
browserAction
popups or tab pages. - The
executeScript
call in dynamically-injected content scripts loaded bytabs.executeScript
, where it hasn't been loaded using acontent_scripts
declaration inmanifest.json
.
So, for example, this manifest.json
code makes the polyfill available to background scripts:
{
// …
"background": {
"scripts": ["browser-polyfill.js", "background.js"]
}
}
Your goal is to ensure that the polyfill executes in your extension before any other scripts that expect the browser.*
API namespace execute.
Note: For more details and information on using the polyfill with a module bundler, see the project's readme on GitHub.
There are other polyfill options. However, at the time of writing, none of the other options provide the coverage of the WebExtension browser API Polyfill. So, where you haven't targeted Firefox as your first choice, your options are to accept the limitations of alternative polyfills, port to Firefox and add cross-browser support, or develop your own polyfill.
API function coverage
The differences in the API functions offered in each of the main browsers fall into three broad categories:
- Lack of support for an entire function.
For example, at the time of writing, Edge didn't support the
browserSettings
function. - Variations in the support for features within a function.
For example, at the time of writing, Firefox doesn't support the notification function method
notifications.onButtonClicked
, while Firefox is the only browser that supportsnotifications.onShown
. - Proprietary functions, supporting browser-specific features.
For example, at the time of writing, containers was a Firefox-specific feature supported by the
contextualIdentities
function.
Details about the support for the extension APIs among the main browsers and Firefox for Android and Safari on iOS can be found on the Mozilla Developer Network Browser support for JavaScript APIs page. Browser compatibility information is also included with each function and its methods, types, and events in the Mozilla Developer Network JavaScript APIs reference pages.
Handling API differences
A simple approach to addressing API differences is to limit the functions used in your extension to functions that offer the same functionality across your range of targeted browsers. In practice, this approach is likely to be too restrictive for most extensions.
Instead, where there are differences among the APIs, you should either offer alternative implementations or fallback functionality. (Remember: you may also need to do this to allow for differences in API support between versions of the same browser.)
Using runtime checks on the availability of a function's features is the recommended approach to implementing alternative or fallback functionality. The benefit of performing a runtime check is that you don't need to update and redistribute the extension to take advantage of a function when it becomes available.
The following code enables you to perform a runtime check:
if (typeof fn === "function") {
// safe to use the function
}
Content script execution contexts
Content scripts can access and modify the page's DOM, just as page scripts can. They can also see any changes made to the DOM by page scripts. However, content scripts get a "clean" view of the DOM.
Firefox and Chrome use fundamentally different approaches to handle this behavior: in Firefox it's called Xray vision, while Chrome uses isolated worlds. For more details, see Content script environment section of the Content scripts concept article.
However, Firefox provides some APIs that enable content scripts to access JavaScript objects created by page scripts and to expose their JavaScript objects to page scripts. See Sharing objects with page scripts for details.
There are also differences between the Content Security Policy (CSP) for content scripts.
Background page and extension service worker
As part of its implementation of Manifest V3, Chrome replaced background pages with extension service workers. Firefox retains the use of background pages, while Safari supports background pages and service workers.
For more information, see the browser support section on the "background"
manifest key page. This includes an example of how to implement a cross-browser script.
Manifest keys
The differences in the manifest.json
file keys supported by the main browsers fall broadly into three categories:
- Extension information attributes.
For example, at the time of writing, Firefox and Opera include the
developer
key for details about the developer of the extension. - Extension features.
For example, at the time of writing, Chrome did not support the
browser_specific_settings
key. - Key optionality.
At the time of writing, generally, only
"manifest_version"
,"version"
, and"name"
are mandatory keys.
Browser compatibility information is included with each key in the Mozilla Developer Network manifest.json
key reference pages.
As manifest.json
files tend to change little—except for release numbers, which may differ between the various browsers—creating and editing a static version for each browser is usually the simplest approach.
Extension packaging
Packaging an extension for distribution through the browser extension stores is relatively straightforward. Firefox, Chrome, Edge, and Opera all use a simple zip format that requires the manifest.json
file to be at the root of the zip package. Safari requires extensions to be packaged in a similar way to apps.
For details on packaging, refer to the guidance on the respective extension's developer portals.
Extension publishing
Each of the major browsers maintains browser extension stores. Each store also reviews your extension to check for security vulnerabilities.
As a consequence, you need to approach adding and updating your extension for each store separately. In some cases, you can upload your extension using a utility.
This table summarizes the approach and features of each store:
Browser | Registration fee | Upload utility | Pre-publication review process | Account two-factor authentication |
---|---|---|---|---|
Chrome |
Yes |
Yes |
Automatic, less than an hour |
Yes |
Edge |
No |
No |
No SLA provided |
Yes |
Firefox |
No |
Automatic, a few seconds. A manual review of the extension takes place after publication, which may result in the extension being suspended where issues that need fixing are found. |
Yes |
|
Opera |
No |
No |
Manual, no SLA provided |
No |
Safari |
Yes |
No |
Yes with, according to Apple, on average, 50% of apps reviewed in 24 hours and over 90% reviewed in 48 hours. |
Yes |
Other considerations
Version numbering
The Firefox, Chrome, and Edge stores require that each uploaded version has a different version number. This means you cannot revert to an earlier version number if you come across issues in a release.
Conclusion
When approaching a cross-platform extension development, the differences between extension API implementations can be addressed by targeting Firefox and using the WebExtension browser API Polyfill.
The bulk of your cross-platform work is likely to focus on handling variations among the API features supported by the main browsers. You may also need to account for differences between the content script and background script implementations. Creating your manifest.json
files should be relatively straightforward and something you can do manually. You then need to account for the variations in the processes for submitting to each extension store.
Following the advice in this article, you should be able to create an extension that works well on all of the four main browsers, enabling you to deliver your extension features to more people.