Difference between revisions of "HTML5 Performance"

From W3C Wiki
Jump to: navigation, search
(Identifying the performance gaps)
(Identifying the performance gaps)
Line 36: Line 36:
 
** Developers coming from web 1.0 backgrounds with rudimentary programming skills who cargo-cult their AJAX apps with resulting poor performance.
 
** Developers coming from web 1.0 backgrounds with rudimentary programming skills who cargo-cult their AJAX apps with resulting poor performance.
 
** Some new low end phones have terrible browser performance across everything
 
** Some new low end phones have terrible browser performance across everything
** The browser on the brand new <censored> is very sluggish doing content animation in general.
+
** '''The browser on this brand new high-end phone is very sluggish doing content animation in general.'''
 
* Are there common patterns/usages where performance becomes an issue?
 
* Are there common patterns/usages where performance becomes an issue?
 
* Are performance gaps similar on desktop and mobile? Are those gaps specific to an implementation, set of devices, or OS?
 
* Are performance gaps similar on desktop and mobile? Are those gaps specific to an implementation, set of devices, or OS?

Revision as of 21:29, 28 March 2013

Getting involved

Join the public-html5-performance@w3.org mailing list.

The Problem

Problem Definition

W3C works on individual components of the web architecture, driven by the functional needs of the W3C participants. This particular focus may lead us to miss system-level considerations such as the overall performance of typical implementations. While performance suffers for a variety of reasons, Web standards could be part of the problem. In the past, when we looked at performance we instrumented the architecture to allow performance to be measured. This is quite different from a system level focus on addressing and improving performance.

As we mature to take a platform level view of our work, we should address performance issues to grow the acceptance of our standards.

What are the indicators that show that there is or is not a problem?

Every week, there is a news article appearing indicating performance issues with HTML5. Recent examples are:

What are the risks?

Any development platform critically needs developers; developers are less likely to pick the Open Web Platform to develop their applications if there are performance issues with it.

Is the problem real?

  • Do we have examples of slow Web applications?
  • Should we and can we compare the performance of native applications with Web applications?

Identifying the performance gaps

  • Where are the bottlenecks?
    • Poor, incomplete or simply missing GPU acceleration for CSS effects
    • Slow DOM interaction performance
    • Crowded main threads that result in stutter for things like fast infinite scrolls
    • Developers coming from web 1.0 backgrounds with rudimentary programming skills who cargo-cult their AJAX apps with resulting poor performance.
    • Some new low end phones have terrible browser performance across everything
    • The browser on this brand new high-end phone is very sluggish doing content animation in general.
  • Are there common patterns/usages where performance becomes an issue?
  • Are performance gaps similar on desktop and mobile? Are those gaps specific to an implementation, set of devices, or OS?

Solutions?

  • Should we advise groups to include a performance considerations section in their specifications?
  • Should we create a "Performance Interest Group", similar to our Privacy Interest Group, to provides guidelines and advice for addressing performance in standards development.?
  • Should we produce more guidelines/best practices?
    • Within the Web Performance Working Group?
    • On webplatform.org?
    • Others?