Closing the loop on Google and webperf

Closing the loop on Google and webperf

You know I’m not one of those people who usually mumble their disagreement in the dark side of the room. That’s a good thing; the world evolves in large part thanks to constructive feedback.

When you’re such a kind of person, it is important to acknowledge when the world has changed and your feedback was taken into account. Today is such a day: I saw web performance talks from Google who made me real happy.

Performance tips that are not new traps you’ll fall into

For a long time, Google web performance talks were made in an ivory tower. Next to great blog posts doing deep dives into a particular topic of the web performance world, talks usually were like movies with tremendous potential, but directed completely wrong.

Using magical hacks that the speakers often barely took time to explain, tips were given to developers based on how to solve a particular performance issue. Remember those ugly “-webkit-transform: translateZ(0)”? When pointed out that their talk didn’t discuss the tradeoffs of the hack and would result in developers actually making worse websites on average, the response was less than stellar. I was ultimately proven right by facts, though…

Fast forward to Google IO 2016. I haven’t been paying attention to most Google perf talks in the past two years due to my general disappointment with those; barely learning anything and ending up angry at the miniguns Google handed to people who seemed to have until then learned to use a beautiful water gun (without telling them not to fire on real people). It was a mistake. I watched a bunch of talks from this year’s Google IO, in part because I have been sick and didn’t have anything much better to do. And I’m glad I did.

High Performance Web UI

If you didn’t watch it already, I cannot not recommend you to watch High performance web user interfaces from Paul Lewis. There were other very good talks at Google IO (Jake Archibald and Chris Wilson close my personal top 3 for web devs) but Paul Lewis did a marvelous job with this talk. Not that he introduced any major technological break-through, no, really not. It introduced a mindset.

How do you approach animations when building with perf in mind. This talk summarizes in 35 short minutes the know-how people like me naturally apply when thinking about making animations for a website. Instead of analyzing bad patterns and trying to fix those, this talk is all about getting it right from the ground up. And even though I learnt nothing new from the technical point of view during this talk, I was delighted to have this thought process formalized, which will enable me to teach it and, better, reason about it in the future instead of letting my intuition and experience do the work alone. It made me a better web dev, and so will the other devs that will be watching too.

For the first time, I left a Google perf talk wishing it would have told people more, not less. Why not speaking about “transform-origin” and “clip”? The final animation demoed clearly makes use of those, not what the talk mentioned. But then again, the talk was only 40 minutes long. I’m sure those topics are or will be discussed in the series that seems linked to it.

In the mean time, every feature that was introduced, was introduced properly, talking about the tradeoffs, how to use it without abusing it. A major breakthrough.

Marvelous about one, but this translates to the other

Talking more broadly about other talks like the Service Worker and Progressive Web Apps ones, I felt in general a clear shift towards helping people build great stuff, versus getting people to fix bad stuff, and I think the Web Perf PR team at Google finally nailed it.

I mean it, those talks were great, focused properly, and the actors were as good as they always were (which is a humble way is to mean awesome).

So, with that said, thank you.

Published on 2016-05-21 under ALL, WEB, PERF, GOOGLE

Comments

A javascript file is loading the comments.