Profile cover photo
Profile photo
Jake Archibald
6,510 followers -
Developer Advocate working at Google
Developer Advocate working at Google

6,510 followers
About
Jake Archibald's posts

Post has shared content
HTTP 203 Podcast!

That's right, you can now get a podcast of +Jake Archibald and me talking about the web! We're just waiting for it to get listed all around the place, but most apps let you add it via a feed URL.... so go go go!

Post has attachment
I had a responsive images epiphany and I wrote it all down before I forgot everything

Post has attachment

Post has attachment
The web must keep developing! My response to ppk suggesting the web should stop adding new features for a year:

Post has attachment
It's a new HTTP 203! +Paul Lewis & I discuss the benefits of progressive loading, and how JavaScript frameworks often get in the way.

Also, I can't speak.

Post has attachment
It's another HTTP 203! This time, I think web security is an important and good thing, whereas +Paul Lewis doesn't because blah blah I wasn't even listening tbh

Post has attachment
It's an all new HTTP 203! +Paul Lewis is worried that the complexity of build tools are creating a steep "on ramp" for the web. Meanwhile, I've ruined a pair of earphones.

Post has attachment
Here's a little video (10 mins) where I look at the performance difference between client rendering, server rendering, and serviceworker rendering.

Post has attachment
There was some confusion over the fetch API, with some claiming it's "high level". It isn't, let's take a look at some of the low-level features.

http://jakearchibald.com/2015/thats-so-fetch/

I've been micro-optimising the performance of https://jakearchibald.github.io/svgomg/ - since the ServiceWorker makes it more likely to have a full cache, I've been looking at time-to-first-render in the full-cache case.

I have some inline CSS & html for first render, then lazy-load in the CSS and <script async> the javascript. However, it was taking 700ms to get from page received to first render on a Nexus 5.

Because the CSS and scripts were cached, they were arriving before first render, that meant the browser was executing/layouting/rendering their results before first render.

The answer? requestAnimationFrame. I was already loading the CSS via JS, but now I'm loading the additional scripts that way too, wrapped in requestAnimationFrame https://github.com/jakearchibald/svgomg/blob/e3dd1ed876ff066bad90a90b63bd319d08e76858/src/index.html#L213

Page-received to first-render on a Nexus 5 went from 700ms to 160ms.
Wait while more posts are being loaded