robcee.net

I sat in on three sessions at Google IO 2013 yesterday.

Memory Lane with Chrome Devtools and GMail was the first.

The presenters showed off their Heap Tracking Profiler and Memory Tracking tool in the Timeline and explained how to use them to track down a leaky DOM node. It was a practical application of how to use a developer tool to solve a particular problem.

One interesting takeaway that surprised the presenters during their research: Always allocating more memory (caching) as a way to improve performance in a large application like Gmail is not a panacea for slow performance. Having a large heap space actually slows down the garbage collector and your performance suffers. It’s a fine balance.

What surprised me was how they analyzed their problem by tracking a user with a known high memory problem for three days. The Google team constantly monitors their apps’ performance via the window.performance API and can single out hotspots in the population.

The next talk I sat in on was about Chrome Apps. The presenter, Erik Kay showed off some of the “Immersive” experiences of Chrome Apps and the different ways they could interact with the hardware on the Chrome Book. The talk included a demo of a small thermal printer being hooked up and controlled over USB which garnered some applause.

The Chrome Web Store lets you buy apps for Chrome.

The only real mention of Android was that they were using PhoneGap and Cordova to provide their compatibility layer. Same for IOS. There will be compatibility issues with deploying on iOS but it seems surprising that they would pursue this completely separate technology for Android. Surely they could ship a full version of the Chrome Runtime and deal with hardware incompatibilities directly.

The questions from the room were interesting. One man (not a Mozillian) asked about WebRTC compatibility across the different platforms, pointedly repeating the question of whether or not he’d be able to use WebRTC in an app on iOS. Only when their WebView supports it.

Another man asked something about interoperability between B2G and ChromeRT. Erik said that there is “no forcing function yet to drive standardization”.

I think my biggest takeaway from this talk was that people wearing Google Glass look like dorks.

My second biggest takeaway was that I was very surprised that there was zero mention of the Google Play Store for Chrome Apps.

Last talk I attended was a Fireside Chat with the Blink Team. While I was expecting an actual fire and was disappointed there wasn’t one, the team bravely took questions from an audience confused about feature-detection, unprefixed CSS and market fragmentation.

Dan Buchner asked the panel something about standardization and I felt a little badly for the Blink team who had a whole chunk of slides talking about how they’re going to be good citizens. (If you want to participate, you should join blink-dev@chromium.org.)

I was interested in ChromeStatus.com/features which shows a spreadsheet of features in-progress. Time will tell how their Intent to Implement and Intent to Ship broadcasting will work from an Open Source point-of-view, but they are currently claiming that a third of their intents to implement are coming from outside of Google.

I wanted to meet Paul Irish after the talk but Steven Shankland showed up and pushed me out of the way. When he was done I did get to meet him, but I think Buchner had made him angry or something. Maybe he was just tired. I dunno.