Leave a Reply

29 Comments on "Future App Model: Advanced Service Worker (Chrome Dev Summit 2016)"

Notify of
avatar

Lars Rye Jeppesen
Guest
Lars Rye Jeppesen
12 days 12 hours ago

Polymorphic webapps kinda solve the spa issues, right? Also, breaking an spa into lazy/preloaded modules makes 1st paint insanely fast. See Angular2 router for example.

Jake Archibald
Guest
Jake Archibald
11 days 22 hours ago
"being able to compose your app with components" – I'm not entirely sure what this means, but web components work fine on non-SPAs."great separation of content and site/layout" – HTML and CSS provides this separation"components that are permanent on the site don't have to re-render on each page load" – right, but I showed with both wiki-offline and github that this doesn't always mean it's faster."I'm not sure I understand how Github managed to do their Spa that slow" – I show why this happens in the talk. A lack of streaming means the browser has to fetch everything before… Read more »
Lars Rye Jeppesen
Guest
Lars Rye Jeppesen
11 days 22 hours ago
Hi Jake, – while you certainly had some great arguments, I'm not sure I can follow why suddenly server side rendering are better than spas.Wouldn't you loose things like:- being able to compose your app with components- great separation of content and site/layout- components that are permanent on the site (settings, navigation etc) don't have to re-render on each page loadI'm not sure I understand how Github managed to do their Spa that slow. In my experience, you can optimize an spa to have first render in < 500 ms by only sending minimal payload for 1st paint, then lazyload/prefetch… Read more »
Jake Archibald
Guest
Jake Archibald
11 days 22 hours ago

If this solution leaves the user with a load of buttons etc that don't work (until the script finally kicks in), it isn't a good solution. Also, once the SPA does kick in, it can be slower when it comes to navigating. See the real-world github example in the talk.

Johannes Ewald
Guest
Johannes Ewald
12 days 15 hours ago

Jake wears sexy socks:

Alastair Taft
Guest
Alastair Taft
14 days 2 hours ago

At 33:45 that's a non starter for me, by waiting 1000 milliseconds the animation is non-interruptible. So if the user hit back after 500 milliseconds, it would either break or appear laggy. Laggy animations are worse than no animations. Checkout 'react-motion', there's no time limits, I think it's a much better way to approach animations.

Alastair Taft
Guest
Alastair Taft
12 days 11 hours ago

I just meant if you have to calculate any sort of duration for an animation it runs counter intuitive to animations imo. Really I was nit picking a small detail from an excellent talk, cheers.

Jake Archibald
Guest
Jake Archibald
12 days 22 hours ago

I think there's some confusion or crossed wires here. In the example I showed the fade transition is 1 second long, but that's only an example. You could calculate these numbers however you want. You could for instance say "if the document has been active less than 2 seconds, don't transition", or calculate durations based on distance – you're writing the script, you can do what you want.What to do if a navigation is triggered during a navigation transition is an interesting question, I've filed github.com/jakearchibald/navigation-transitions/issues/12 so we come up with a solution.

Hiren Patel
Guest
Hiren Patel
15 days 15 hours ago

Fantastic talk 🙂

b j
Guest
b j
19 days 18 hours ago

my bugbear with service worker is the need for https

b j
Guest
b j
11 days 18 hours ago

fair enough and thanks for the heads up.

Lars Rye Jeppesen
Guest
Lars Rye Jeppesen
11 days 19 hours ago

b j well companies like Google support it, so don't think there are issues I use it and it rocks

b j
Guest
b j
11 days 19 hours ago

thanks – however in this case – one could say that running "certbot" on the server creates a larger potential security compromise than the ability to cache squirrel pics using service worker via http.

Lars Rye Jeppesen
Guest
Lars Rye Jeppesen
11 days 22 hours ago

letsencrypt.org It will change your life

Simon Hamp
Guest
Simon Hamp
19 days 23 hours ago

Seriously brilliant talk.

Tom
Guest
Tom
21 days 47 minutes ago

are you holding a Wii remote in your hand?

Tom
Guest
Tom
18 days 22 hours ago

Jake Archibald awesome! 😀

Jake Archibald
Guest
Jake Archibald
19 days 2 hours ago

Tom cheaper to use the Bluetooth device I already own than buy a dedicated clicker 😁

Tom
Guest
Tom
19 days 16 hours ago

Jake Archibald but but why? 😀 are you using the infrared to control your PC?

Jake Archibald
Guest
Jake Archibald
20 days 17 hours ago

Tom yep!

Zane Hannan
Guest
Zane Hannan
21 days 11 hours ago

Neat.Also, I like the idea of using a Wiimote as a cheaper solution to the clicker. It looks neater too, can anyone explain it?

Zane Hannan
Guest
Zane Hannan
18 days 6 hours ago

Thanks!

Jake Archibald
Guest
Jake Archibald
20 days 17 hours ago

mattrick d-pad is mapped to arrows, but trigger button is mapped to space, which I use as "move forward with animation". i'm using darwiimote, but I think there are better tools, and yeah glovepie is ace on Windows

mattrick
Guest
mattrick
21 days 7 hours ago

I think the way it works is that he's mapped the D-pad on the Wiimote to the arrow keys on his computer using something like GlovePIE (Google it). The Wiimote can connect to computers using Bluetooth (its fairly easy to do, Google it ;D).

Bogdan Mitrovic
Guest
Bogdan Mitrovic
21 days 8 hours ago

Zane Hannan didnt even notice that lol

Gabriel Kwiecinski Antunes
Guest
Gabriel Kwiecinski Antunes
21 days 17 hours ago
Gabriel Kwiecinski Antunes
Guest
Gabriel Kwiecinski Antunes
21 days 17 hours ago

You did it. You actually fucking did it. Look at you all dressed up talking about V1.

wpDiscuz