What is the difference between HTML5 AppCache and the normal browser cache?

Question

I don't understand the point of the HTML5 AppCache. We already have a normal cache. If you visit a website the first time it'll already cache all the assets. What extra value does the AppCache provide? Is it just a list of files so that the browser knows what assets to download, even if they're not referenced by the HTML right now? Does the browser make sure that the caching is "all-or-nothing", i.e. does it ensure that everything referenced by the manifest is cached, or nothing at all?

Answer 1

I think the point you're missing is that AppCache is specifically designed to allow web apps (and web sites) to be made available offline, though the same speed benefits which the normal browser cache provides, when the user is online, are also provided by AppCache.

The key difference with the browser cache is that you can specify all the assets the browser should cache in a manifest file (conceivably your entire site) whereas the browser cache will only store the pages (and associated assets) you have actually visited.

I'm no expert on the AppCache, but I do know it is not without its problems. There's a really good article here from a chap who used AppCache to allow parts of his mobile site to be available offline. It includes some rationale on their decision to use it and a number of gotchas they encountered in doing so.

This HTML5 Rocks article on the subject also has some good information.

Answer 2

AppCache actually uses the browser cache in support of its operation. It is the browser equivalent of downloading an app to run locally.

The first time a user visits the page, the resources of that page will be loaded from the server and stored in the normal cache. If the page specifies an appcache manifest, the browser will download the manifest and fetch all the resources in there (even if they do not appear on the page that embedded the manifest). These are then stored in appcache.

The second time a user visits the page, the browser will check its appcache. If an entry exists for that URL, it will load the page from appcache instead of from the server, based on the rules specified in the manifest (the manifest can mark some resources explicitly as fetched from the network).

After the browser loads the page from appcache, it will contact the server to see if there is an updated manifest. If the manifest is updated, it will fetch the resources from the manifest. These fetches are done using normal browser cache rules, so some of these resources may actually end up being fetched from the regular browser cache instead of from the server (this allows you to do differential updates when using appcache to develop offline apps). The new version of the appcache is kept separate from the old version. After the new version is fetched the user keeps interacting with the resources from the old version until they refresh the main page, after which the new version is loaded and the old is discarded.

Another important point is that appcache has different rules for when resources are discarded. Appcache basically never discards the latest set of resources, and caches them as a whole. To prevent abuse it enforces storage limits (sometimes as little as 5 MB) of how big a site's cache can be. By contrast, the browser cache has no per-site limits, but will discard individual resources from a site if the global cache limits are reached.