Yahoo and their development team in particular deserve kudos much more often than they seem to get. PR-wise, they get elbowed out by Google, FB, etc. when in fact their open-source tools like YUI are often very well made, well tested (if it's good enough for the Yahoo front page, it's probably pretty good), and most importantly for fellow developers, well documented. Thanks Yahoo!

Oh, and I forgot to mention YSlow - as far as I can tell, that Firefox plugin was available years before anything else like it (e.g. Chrome's performance profiler). Not only did it work and work well, but the Yahoo! devs released lots of documentation and videos around how to improve site performance, i.e. free education from people who dealt with massive scale daily.

Yeah, used yslow recently and its got a lot of good stuff. I'd never even heard of an etag before. Or Chrome's profiler, for that matter. Thanks!

Completely agree. Personally, I've been using the javascript testing framework from YUI 3 and it's been great so far. It's pretty easy to setup and I can even have the test output as a JUnit xml file. Did someone say Hudson integration?

http://developer.yahoo.com/yui/3/test/#test-reporting

Steve "High Performance" Souders jumped ship and went to Google though.

Steve definitely kicked "high performance" into overdrive at Y!, but he's one of many people who have contributed from Yahoo:

Stoyan Stefanov write weekly about performance on his blog http://www.phpied.com/ and actually just released "Web Testing Framework" at Velocity which is a YSlow plugin.

Nicholas Zakas wrote "High Performance JavaScript" (http://www.amazon.com/Performance-JavaScript-Faster-Applicat...) and does a ton of stuff for the Yahoo frontpage.

Phil Tellis (author of Boomerang) regularly blogs on performance on http://bluesmoon.info

I (sh1mmer) have done a bunch of performance related stuff with Aristus (now at FB) and I'm working on how network, consumer hardware and other holistic factors affect web performance. Expect some DNS related stuff soon.

It's worth noting that this won't help you with timing the first page loaded, just the ones after that. Impressive nonetheless.

That's a really good point, it's actually a constraint of the architecture of the web. This framework is based on our own research of the best way to reliably measure page performance is across page loads.

Boomerang itself is actually a generic beaconing framework so you can use it to beacon anything but it also has a lot of bootstrap code as well for performance testing.

Phil has done some great documentation on how to do beaconing right to go with the framework and you should expect more of that.

I'm hoping that browsers will soon support WebTiming: http://dev.w3.org/2006/webapi/WebTiming/ We'll add support into boomerang for those APIs

However, until then, you can still measure the time from first-byte to onload by starting a timer at the top of your page. It's crude, but it works.

The IE9 preview-release they talked about at Velocity does and it sounds like other browser manufacturers are in given the discussion on the browser panel, but as usual that means it's 2 years before it's useful.

Ok, I've added the WebTiming API checks to boomerang.

you just scored high on the DeveloperTiming API.