MDN has a standard format for compatibility tables for our open web documentation; that is, documentation of technologies such as the DOM, HTML, CSS, JavaScript, SVG, and so forth, that are shared across all browsers. This article covers how to use our features to add compatibility data to MDN pages.

Important: The way the data is generated has changed. Historically our tables were inserted on the page and the data populated manually. This is inefficient, makes maintenance difficult and causes the data to be inflexible. As a result, we are migrating our browser compat data to be stored in a data repo (see https://github.com/mdn/browser-compat-data) and generating the tables programmatically.

In this guide, we document the new way to add compat data to MDN, but we've still kept documentation covering the old way, as you'll see manual tables on MDN for a while. If you need to see the old documentation, check out our Old compatibility tables article.

Note: If you need any help with any steps of this guide, you are welcome to contact us at the MDN discussion forum.

How to access the data repo

The data is stored in a GitHub repo — see https://github.com/mdn/browser-compat-data. To access it, you need to get a GitHub account, fork the browser-compat-data repo over to your own account, then clone your fork onto your local machine.

Choosing a feature to add data for

First of all, find a feature that you want to add browser-compat-data for. This could be an HTML element, CSS property, JS language feature or JS API interface, for example. We would like to encourage you to work on API features, as we already have people working on HTML, JS, and CSS. You can find the status of features that need their data adding to the repo on our Browser Compat Data migration spreadsheet.

The process for using this is as follows:

  1. Go ahead and choose a feature that is not already being worked on or complete. Put your name in the "Who" column, preferably along with your MDN username so we can find your email address and contact you if we need to.
  2. If the feature you want to work on is not already listed in the spreadsheet, add rows for it in an appropriate place, copying the format that is already there. You should also use the same granularity (e.g. per element for HTML, per property or selector for CSS, per object for JS, per interface for APIs).
  3. Once you've started work on adding the data, put the status to "In progress".
  4. Once you've added the data and submitted a pull request to the main repo, put the status to "PR done".
  5. As your data is merged to the repo, then added to the npm package, update the status as necessary.
  6. Once you've updated the documentation page(s) for your feature to use the new macro to dynamically generate the appropriate data table on each page, set the status to "Article updated". At this point, you are done.

Preparing to add the data

Before adding some new data, you should make sure that your fork is up-to-date with the main repo (it contains the same content), create a new branch inside your fork to contain your additions, then pull that branch into your local clone so you can start working inside it:

Let's look at a simple way to make sure your fork is to-to-date is as follows:

Adding the main browser-compat-data repo as a remote

Go to your local clone of your fork in your terminal/command line, and add a remote pointing to the main (upstream) repo like so (you only need to do this once):

git remote add upstream https://github.com/mdn/browser-compat-data.git

If you are unsure whether you've done this, you can check what remotes your repo has using

git remote -v

Updating your fork with the remote's content

Now, whenever you want to update your fork, you can do so by:

  1. Making sure you are in the master branch:

    git checkout master
  2. fetching the up-to-date repo contents using the following:

    git fetch upstream
  3. rebasing the contents of your master with the main repo's contents:

    git rebase upstream/master
  4. pushing these updates back to your remote fork using this:

    git push -f

Creating a new branch to do your work in

Next, go to your remote fork (it will be at https://github.com/your-username/browser-compat-data) and create a new branch to store your changes for this data addition. This can be done by:

  1. Clicking on the "Branch: Master" button.
  2. Entering a new branch name into the "Find or create a branch..." text field.
  3. Pressing the resulting "Create branch name-of-branch from Master" button.

For example, if you were wanting to add data for the WebVR API, you'd create a branch called something like "webvr".

Switching to the new branch

At this point, go back to your terminal/command line, and update your fork's local clone to include your new branch using the following command:

git pull

Now switch to your new branch using this:

git checkout -b name-of-branch

You should now be ready to start adding your data!

Adding the data

To add the data, you need to create a new file or files to store your compat data in. The files you need to create differ, depending on what technology you are working on:

  • HTML: One file per HTML element, contained in browser-compat-data/html/elements. The file should be called the name of the element, all in lower case, e.g. div.json.
  • CSS: One file per CSS property or selector, contained in the appropriate directory (see browser-compat-data/css). The file should be called the name of the feature, all in lower case, e.g. background-color.json, or hover.json.
  • JS: One file per JS object, contained in browser-compat-data/javascript/builtins. The file should be called the exact name of the object, with the casing preserved, e.g. Date.json or InternalError.json.
  • APIs: One file per interface contained in the API, put in browser-compat-data/api. Each file should be called the exact name of the interface, with the casing preserved, e.g. The WebVR API has VRDisplay.json, VRDisplayCapabilities.json, etc.

Note: You'll notice that the repo also contains data for Browser Extensions and HTTP. These data sets are basically finished as they stand, but more features may need to be added in the future.

Each file you create has to follow the pattern defined in the schema contained within our repo; you can see the detailed schema description here.

Basic compat data structure

Let's look at an example. CSS property JSON files for example need the following basic structure:

{
  "css": {
    "properties": {
      "border-width": {
        "__compat": {
          ...
        }
      }
    }
  }
}

You have the css object, inside of which is a properties object. Inside the properties object, you need one member for each of the specific features you want to define the compat data for. Each of these members has a __compat member, inside of which the actual data goes.

The above data is found in the browser-width.json file — compare this to the rendered browser-width support table on MDN.

Other types of features work in the same way, but with different object names:

  • CSS selectors work in basically the same way as CSS properties, except that the top-level object structure is css.selectors instead of css.properties. See cue.json for an example.
  • HTML data works in basically the same way, except that the top-level object structure is html.elements. See article.json for an example.
  • The top level object structure for JS built-in objects is javascript.builtins; see Array.json for an example.

In HTML, CSS, and JS pages, you'll normally only need one feature. API interfaces work slightly differently — they always have multiple sub-features (see Sub-features, below).

Basic structure inside a feature

Inside a feature __compat member, you need to include the following members:

  • mdn_url: Contains the URL of the reference page for this feature on MDN. Note that this needs to be written without the locale directory inside, e.g. /docs/... not /docs/en-US/... (or whatever). This is added in by the macro when the data is put on the page, for localization purposes.
  • support: Contains members representing the browser support information for this feature in all the different browsers we want to report.
  • status: Contains members reporting the standards track status of this feature.

The names of the browser members are defined in the schema (see Browser identifiers). You should use the full list of currently defined identifiers. If you wish to add another browser, talk to us first, as this could have a wide-ranging impact and should not be done without careful thought.

In a basic browser compat data file, you'll only need to include "version_added" inside the browser identifier members (we'll cover Advanced cases later on). The different values you might want to include are as follows:

  • A version number: If you know the exact version in which a browser started to support your feature, use a string representing the number, e.g. "47".
  • true: If a browser supports a feature but you don't know the exact version number, use the value true. This equivalent to the {{CompatVersionUnknown}} macro call in the old manual tables.
  • false: If a browser does not support a feature, use the value false. This is equivalent to the the {{CompatNo}} macro call in the old manual tables.
  • null: If you don't know whether a browser supports a feature or not, use the value null. This is equivalent to the {{CompatUnknown}} macro call in the old manual tables.

Inside the status member, you'll include three submembers:

  •  "experimental": This should be set to true if the feature is experimental, or false otherwise.
  • "standard_track": This should be set to true if a feature is on some kind of standards track (most commonly W3C/WHATWG, but there are also other standards efforts such as Khronos, TC39, etc.) or false otherwise.
  • "deprecated": This should be set to true if the feature is deprecated, or false otherwise.

The feature data for border-width (also see border-width.json) is shown below as an example:

"__compat": {
  "mdn_url": "https://developer.mozilla.org/docs/Web/CSS/border-width",
  "support": {
    "chrome": {
      "version_added": "1"
    },
    "webview_android": {
      "version_added": "2"
    },
    "edge": {
      "version_added": true
    },
    "edge_mobile": {
      "version_added": true
    },
    "firefox": {
      "version_added": "1"
    },
    "firefox_android": {
      "version_added": "1"
    },
    "ie": {
      "version_added": "4"
    },
    "ie_mobile": {
      "version_added": "6"
    },
    "opera": {
      "version_added": "3.5"
    },
    "opera_android": {
      "version_added": "11"
    },
    "safari": {
      "version_added": "1"
    },
    "safari_ios": {
      "version_added": "3"
    }
  },
  "status": {
    "experimental": false,
    "standard_track": true,
    "deprecated": false
  }
}

Adding a description

There is a fourth, optional, member that can go inside the __compat member — description. This can be used to include a human-readable description of the feature. You should only include this if it is hard to see what the feature is from glancing at the data. For example, it might not be that obvious what a constructor is from looking at the data structure, so you can include a description like so:

{
  "api": {
    "AbortController": {
      "__compat": {
        ...
      },
      "AbortController": {
        "__compat": {
          "mdn_url": "https://developer.mozilla.org/docs/Web/API/AbortController/AbortController",
          "description": "<code>AbortController()</code> constructor",
          "support": {
            ...
          }
        }
      }

      ... etc.
    }
  }
}

Sub-features

In a page where the compat table has more than one row, you'll need multiple subfeatures inside each feature to define the information for each row. This can happen, for example, when you've got the basic support for a feature stored in one row, but then the feature also has a new property or value type that was addded much later in the specification's life and is only supported in a couple of browsers.

As an example, see the compat data and corresponding MDN page for the background-color property. The basic support exists inside the __compat object as explained above, then you have an additional row for browsers' support for "alpha channel for hex values", which contains its own __compat object.

{
  "css": {
    "properties": {
      "background-color": {
        "__compat": {
          ...
        },
        "alpha_ch_for_hex": {
          "__compat": {
            ...
          },
        }
      }
    }
  }
}

For an API, you've got the top two levels defined as api.name-of-the-interface, then a top-level __compat section to define the overall browser compatibility of the interface, then a sub-feature for each of the methods, properties, and constructors contained inside the interface. The basic structure looks like this:

{
  "api": {
    "VRDisplay": {
      "__compat": {
        ...
      },
      "cancelAnimationFrame": {
        "__compat": {
          ...
        }
      },
      "capabilities": {
        "__compat": {
          ...
        }
      },

      ... etc.
      
    }
  }
}

See VRDisplay.json for a full example.

Adding data: Advanced cases

There are some advanced features that you'll want to include in browser compat data. The aim of this section is to list the most common ones, providing an example of each to show how you can implement them in your own compat data.

Including a footnote

Often compat tables will include footnotes related to certain entries that explain useful details or strange behavior that developers will find useful. As an example, the Chrome Android entry for VRDisplay.capabilities (see also VRDisplay.json)  (at the time of writing) had a footnote "Currently supported only by Google Daydream." To include this in the capabilities data, we added a "notes" submember inside the relevant "chrome_android" submember; it would look like this:

"chrome_android": {
  "version_added": true,
  "notes": "Currently supported only by Google Daydream."
}

Including a vendor prefix

If a feature is supported behind a vendor prefix in one or more browsers, you'll want to make that clear in the browser compat data. imagine you had a feature that was supported with a -moz- prefix in Firefox. To specify this in the compat data, you'd need to add a "prefix" submember inside the relevant "firefox" submember. It would look something like this:

"firefox": {
  "version_added": true,
  "prefix": "-moz-"
}

Including a browser pref/flag

Some features may be supported in a browser, but they are experimental and turned off by default. If a user wants to play with this feature they need to turn it on using a pref/flag.

To represent this in the compat data, you need to add the "flag" submember inside the relevant browser identifier submember. The value of "flag" is an object containing three members:

  • "type": The type of flag or pref this is. The most common value is "preference", which is set inside the browser (for example, using about:config in Firefox, or chrome://flags in Chrome), but you might also sometimes use a value of "compile_flag", which is a preference set when the browser build is compiled.
  • "name": This is a string representing the name of the preference that needs to have a value set on it. For example, "Enable Experimental Web Platform Features" is a preference that exists in Chrome, found in chrome://flags.
  • "value_to_set": This is a string representing the value that needs to be set on the preference, for example "true".

So to add a pref/flag to the Chrome support for a feature, you'd do something like this:

"chrome": {
  "version_added": "50",
  "flag": {
    "type": "preference",
    "name": "Enable Experimental Web Platform Features",
    "value_to_set": "true"
  }
},

Including a version where support was removed

Sometimes a feature will be added in a certain browser version, but then removed again as the feature is deprecated. This can be easily represented using the "version_removed" submember, which takes as its value a string representing the version number it was removed on. For example:

"firefox": {
  "version_added": "35",
  "version_removed": "47",
},

Including multiple support points for the same browser entry

Sometimes you'll want to add multiple support data points for the same browser inside the same feature.

As an example, the text-align-last property (see also text-align-last.json) was added to Chrome in version 35, supported behind a pref.

The support mentioned above was then removed in version 47; also in version 47, support was added for text-align-last enabled by default.

To include both of these data points, you can make the value of the "chrome" submember an array containing two support information objects, rather than just a single support information object:

"chrome": [
  {
    "version_added": "47"
  },
  {
    "version_added": "35",
    "version_removed": "47",
    "flag": {
      "type": "preference",
      "name": "Enable Experimental Web Platform Features",
      "value_to_set": "true"
    }
  }
],

Note: You should put the most current or important support point first in the array — this makes the data easier to read for people who just want to scan it for the latest info.

Including an alternative name

Occasionally browsers will support a feature under a different name to the name defined in its specification. This might be for example because a browser added experimental support for a feature early, and then the name changed before the spec stabilized.

To include such a case in the browser compat data, you can include a support information point that specifies the alternative name inside an "alternative_name" member.

Note: The alternative name might not be an exact alias — it might have differing behaviour to the standard version.

Let's look at an example. The border-top-right-radius property (see also border-top-right-radius.json) was supported in Firefox:

  • From version 4 onwards with the standard name border-top-right-radius.
  • From version 49 onwards with a -webkit- prefix, for browser compatibility purposes.
  • From version 1 onwards with the alternative name -moz-border-radius-topright. Support for this alias was removed in version 12.

To represent this in the data, we used the following JSON:

"firefox": [
  {
    "version_added": "4",
    "notes": "Prior to Firefox 50.0, border styles of rounded corners were always rendered as if <code>border-style</code> was solid. This has been fixed in Firefox 50.0."
  },
  {
    "prefix": "-webkit-",
    "version_added": "49",
    "notes": "From Firefox 44 to 48, the <code>-webkit-</code> prefix was available with the <code>layout.css.prefixes.webkit</code> preference. Starting with Firefox 49, the preference defaults to <code>true</code>."
  },
  {
    "alternative_name": "-moz-border-radius-topright",
    "version_added": "1",
    "version_removed": "12"
  }
],

Pushing a change back to the main repo

Once you are finished with adding your compat data, you should first test it using one or both of the following commands:

  • npm run lint — tests all the compat data to make sure the JSON is valid, and is written in the correct style. It will print out a long list of file names and test results; if an error is found, the linter will throw an error on the file it is found in, giving you useful debugging info like line number, error message, etc.
  • npm run render 'dotted.path.to.feature' — allows you to preview the markup for the compat table for a data file in the repo. As an example, npm run render 'css.properties.background' shows the table markup for the background property.

If it is looking OK, you then need to commit it and push it back up to your remote fork on GitHub. You can do this easily with terminal commands like this:

git add .
git commit -m 'adding compat data for name-of-feature'
git push

Now go to your remote fork (i.e. https://github.com/your-username/browser-compat-data) and you should see information about your push at the top of the files list (under "Your recently pushed branches"). You can create a pull request (starting the process of pushing this to the main repo) by pressing the "Compare & pull request" button, then following the simple prompts on the subsequent screen.

At this point, you just need to wait. A reviewer will review your pull request, and merge it with the main repo, OR request that you make changes. If changes are needed, make the changes and submit again until the PR is accepted.

Inserting the data into MDN pages

Once your new data has been included in the main repo, you can start dynamically generating browser compat tables based on that data on MDN pages using the {{Compat}} macro. This takes a single parameter, the dot notation required to walk down the JSON data and find the object representing the feature you want to generate the compat table for.

Above the macro call, to help other contributors finding their way, you should add a hidden text that is only visible in MDN contributors in edit mode:

<div class="hidden">
<p>The compatibility table on this page is generated from structured data. 
If you'd like to contribute to the data, please check out 
<a href="https://github.com/mdn/browser-compat-data">https://github.com/mdn/browser-compat-data</a> 
and send us a pull request.</p>
</div>

As an example, on the Accept-Charset HTTP header page, the macro call looks like this: {{Compat("http.headers.Accept-Charset")}}. If you look at the accept-charset.json file in the repo, you'll see how this is reflected in the JSON data.

As another example, The compat table for the VRDisplay.capabilities property is generated using {{Compat("api.VRDisplay.capabilities")}}. The macro call generates the following table (and corresponding set of notes):


FeatureChromeEdgeFirefoxInternet ExplorerOperaSafari
Basic support Yes1 2 Yes

553

584

No ? ?
FeatureAndroid webviewChrome for AndroidEdge mobileFirefox for AndroidIE mobileOpera AndroidiOS Safari
Basic support No Yes5 ? ? No ? ?

1. Only works on desktop in an experimental version of Chrome (other builds won't return any devices when Navigator.getVRDisplays() is invoked).

2. This feature is behind the WebVR preference. To change preferences in Chrome, visit chrome://flags.

3. Windows support was enabled in Firefox 55.

4. macOS support was enabled in Firefox 58.

5. Currently supported only by Google Daydream.

Document Tags and Contributors

 Last updated by: Sheppy,