livereload-js

LiveReload.js

What is LiveReload?

LiveReload is a tool for web developers and designers. See livereload.com for more info.

To use LiveReload, you need a client (this script) in your browser and a server running on your development machine.

This repository (livereload.js) implements the client side of the protocol. The client connects to a LiveReload server via web sockets and listens for incoming change notifications. When a CSS or an image file is modified, it is live-refreshed without reloading the page. When any other file is modified, the page is reloaded.

The server notifies the client whenever a change is made. Available servers are:

• LiveReload app for Mac
• rack-livereload
• guard-livereload
• grunt-contrib-watch
• python-livereload
• more available on Google :-)
• you can even write your own; refer to the LiveReload protocol

If you are a web developer looking to use LiveReload, you should refer to your LiveReload server/app/tool's documentation, rather that this repository. You should use the copy of livereload.js script bundled with your server, because it's guaranteed to be compatible, and may be customized for that server.

Most LiveReload server vendors will serve livereload.js on the LiveReload port. When your server is running, you can typically access the script at http://0.0.0.0:35729/livereload.js.

Please read on only if you are:

• using a server that doesn't document the usage of livereload.js
• interested in hacking on livereload.js or want to understand it better
• developing a LiveReload server

What is livereload.js?

This repository contains a JavaScript file implementing the client side of the LiveReload protocol. It gets change notifications from a LiveReload server and applies them to the browser.

If you are developing a LiveReload server, see dist/livereload.js for the latest version built using the sources in this repository. We require LiveReload server vendors to distribute livereload.js as part of their apps or tools.

An old version of this script is also bundled with the LiveReload browser extensions, but it's not getting updated and only serves for compatibility with very old clients.

Features:

• Live CSS reloading
• Full page reloading
• Protocol, WebSocket communication
• CSS @import support
• Live image reloading (<img src="..." />, background-image and border-image properties, both inline and in stylesheets)
• Live, in-browser LESS.js reloading

Would love, but doesn't seem possible:

• live JS reloading

Installing using Bower

This script is published on Bower. (But, to reiterate: the preferred method is to avoid installing it altogether, and instead use the one bundled with your LiveReload server/app/tool.)

Installation:

bower install livereload-js --save-dev

This gives you a component containing a single script file, dist/livereload.js.

Installing using npm and Browserify

Including livereload.js into your Browserify bundle probably makes no sense, because livereload.js isn't something you would ship to production.

But if you insist and you know what you're doing, you can install LiveReload via npm:

npm install livereload-js --save

and then add this to your bundle:

window.LiveReloadOptions = { host: 'localhost' };
require('livereload-js');

Note that livereload-js package uses window and document globals, so won't run under Node.js environment.

The reason you need to specify LiveReloadOptions is that livereload.js won't be able to find its <script> tag and would normally bail out with an error message.

Using livereload.js

This script is meant to be included into the web pages you want to monitor, like this:

<script src="http://localhost:35729/livereload.js"></script>

LiveReload 2 server listens on port 35729 and serves livereload.js over HTTP (besides speaking the web socket protocol on the same port).

A slightly smarter way is to use the host name of the current page, assuming that it is being served from the same computer. This approach enables LiveReload when viewing the web page from other devices on the network:

<script>document.write('<script src="http://'
    + location.host.split(':')[0]
    + ':35729/livereload.js"></'
    + 'script>')</script>

However, since location.host is empty for file: URLs, we need to account for that:

<script>document.write('<script src="http://'
    + (location.host || 'localhost').split(':')[0]
    + ':35729/livereload.js"></'
    + 'script>')</script>

LiveReload.js finds a script tag that includes …/livereload.js and uses it to determine the hostname/port to connect to. It also understands some options from the query string: host, port, snipver, mindelay and maxdelay.

snipver specifies a version of the snippet, so that we can warn when the snippet needs to be updated. The currently recommended snipver is version 1:

<script>document.write('<script src="http://'
    + (location.host || 'localhost').split(':')[0]
    + ':35729/livereload.js?snipver=1"></'
    + 'script>')</script>

Additionally, you might want to specify mindelay and maxdelay, which is minimum and maximum reconnection delay in milliseconds (defaulting to 1000 and 60000).

Alternatively, instead of loading livereload.js from the LiveReload server, you might want to include it from a different URL. In this case, add a host parameter to override the host name. For example:

<script src="https://github.com/livereload/livereload-js/raw/master/dist/livereload.js?host=localhost"></script>

Options

Options can either be specified as query parameters of the <script src="..../livereload.js"> tag's source URL, or as a global window.LiveReloadOptions dictionary. If the dictionary is specified, livereload.js does not even try looking for its <script> tag.

The set of supported options is the same for both methods:

• host: the host that runs a LiveReload server; required if specifying LiveReloadOptions, otherwise will be autodetected as the origin of the <script> tag
• port: optional server port override
• path: optional path to livereload server (default: 'livereload')
• mindelay, maxdelay: range of reconnection delays (if livereload.js cannot connect to the server, it will attempt to reconnect with increasing delays); defaults to 1,000 ms minimum and 60,000 ms maximum
• handshake_timeout: timeout for a protocol handshake to be completed after a connection attempt; mostly only needed if you're running an interactive debugger on your web socket server
• isChromeExtension: reload chrome runtime instead of page when true (default: false)
• reloadMissingCSS: prevent reload of CSS when changed stylesheet isn't found in page (default: true)

Issues & Limitations

Live reloading of imported stylesheets has a 200ms lag. Modifying a CSS @import rule to reference a not-yet-cached file causes WebKit to lose all document styles, so we have to apply a workaround that causes a lag.

Our workaround is to add a temporary <link /> element for the imported stylesheet we're trying to reload, wait 200ms to make sure WebKit loads the new file, then remove <link /> and recreate the @import rule. This prevents a flash of unstyled content. (We also wait 200 more milliseconds and recreate the @import rule again, in case those initial 200ms were not enough.)

Live image reloading is limited to <img src="..." />, background-image and border-image styles. Any other places where images can be mentioned?

Live image reloading is limited to jpg, jpeg, gif, and png extensions. Maybe need to add svg there? Anything else?

Communicating with livereload.js

It is possible to communicate with a running LiveReload script using DOM events:

• fire LiveReloadShutDown event on document to make LiveReload disconnect and go away
• listen for LiveReloadConnect event on document to learn when the connection is established
• listen for LiveReloadDisconnect event on document to learn when the connection is interrupted (or fails to be established)

The LiveReload object is also exposed as window.LiveReload, with LiveReload.disconnect(), LiveReload.connect(), and LiveReload.shutDown() available. However, I'm not yet sure if I want to keep this API, so consider it non-contractual. (And please tell me if you have a use for it!)

Having trouble?

To enable debugging output to console, append ?LR-verbose to your URL.

Hacking on LiveReload.js

Requirements:

• Node.js with npm
• Grunt (npm install grunt-cli)

To install additional prerequisites:

npm install

To build:

grunt build

To run tests:

grunt

Manual testing: open files in test/html/* in various browsers, make some changes and make sure they are applied.

Testing the Browserify usage scenario: grunt browserify:test, then perform manual testing of test/html/browserified/.

Releasing a new version

1. Update the version number in package.json.

2. Run rake version to update the version numbers in all other files, using the one from package.json.

3. Run grunt.

4. Do some manual testing.

5. Tag the version in Git: rake tag then git push --tags.

6. npm publish

License

livereload-js is available under the MIT license. See the LICENSE file for details.

Version history

2.2.1 (Jan 17, 2015)

• npm fix: actually include /lib in the package

2.2.0 (Jan 16, 2015)

• the first version stitched with Browserify; everything seems to work, but 2.1.0 is available just in case
• switched the build system to Grunt

2.1.0 (Jan 16, 2015)

• use case-insensitive matching for rel attribute in <link rel="stylesheet"> tags, to accommodate legacy Rails versions
• avoid usage of console when it's not defined
• some README changes

2.0.8 (May 22, 2012)

• Fix live CSS refresh to work with prefixfree
• Correctly trigger removal of old <link> tags

(older history not recorded)