Like Online!

But not.

Why Offline

Increasing use of websites as applications

Never possible to guarantee connection on the server or client side, regardless of fallbacks

Risk of downtime can restrict uptake when software is mission critical

Developers blood pressure is generally too low anyway

How Offline

Separate components deal with code storage and data storage

Local Storage 
a JavaScript interface for simple string storage 

App Cache
a new standard for caching linked resources and HTML
a really, really annoying one

Local Storage

Really basic string storage:
var hero=localStorage.getItem('morning');
alert(hero); // outputs ben 

We can get more utility with JSON

var rick=JSON.parse(localStorage.getItem('rick'));
alert(rick.drinks[0]); // outputs beer 

Can be expanded using

Storage.prototype.twitter = function() { alert('poopin'); } 

Application Cache

<html lang="en" manifest="offline.appcache">

A single file listing all dependencies.


HTML content is cached by the presence of this file.

Including files in the appcache does not link them in the application. They must still be linked with HTML elements

So Far So Good

We already had some pretty good caching mechanisms with HTTP
 Expires: Thu, 6 Oct 2013 03:14:15 GMT
Modifying a query string on a resource link re-fetches
 <link rel="stylesheet" href="style.css?201308061930"/>

Application Cache will tell you where to go with your smart ass ideas

Where it Goes Wrong

When a page loads, appcache is checked first

HTML and other resources are loaded from the cache

Then, regardless of anything else, the page is rendered

Then the appcache file is re-downloaded from it's original URL

So the page rendered for the user is always their last cached page

Where it Goes Wronger

Because the HTML is rendered from the cache, no changes to query strings or headers will re-download your page

Only changes to the appcache file itself will allow new resources to be downloaded

At which point resources are downloaded again

Or not because these also still follow old caching rules

In Case You Missed It

Only changes to the appcache file itself will work

Changing the URL of the file won't work - because the HTML won't be re-downloaded

Because the appcache re-download uses a regular URL, old caching also applies here

If you set a far future header on your appcache file you will get a migraine
And not be able to update your users' sites

Oh and Also

If something is not included in the appcache file it won't be downloaded, even if included as a resource in the HTML

You can get round this, using a header in the appcache
But it will frustrate you if you forget!

Even better if the appcache file changes and any resource in the new file cannot be downloaded, the whole process fails and the old site is rendered

Good Parts

It does work.


Even though your page does render first, a JavaScript updateready  event will fire if new updates are available - so you can at least do something

Using a build tool like Phing can help alter your appcache file for redownloading, and even help ensure all resources served by your page are included

Thanks for Listening

Slides available online

Follow on Twitter:

See a demo of offline applications tomorrow night at @Manc_JS
7th August. TechHub Manchester. 7pm

Useful links:

The Interactive Bit!

First install Jekyll
 gem install jekyll

Or just
 git clone
Checkout the first step
 git checkout stage-0
In the directory, run
 jekyll serve --watch
And open the directory in your favourite editor (or vim)

Go Eat Pizza First!

Offline Web Applications

By Mike Lehan

Offline Web Applications

A lightning talk at the @PHPNW meetup in Manchester, 6th August 2013

  • 2,910