"We finally managed to do general artificial intelligence on the web" :-) Does it have to be already running in non-web environments?

You are too kind <3

I think you are being purposely inflammatory here. We have been working hard on bringing capabilities to the web platform, together with the other browser vendors.

I’m curious what makes you think we’d waste this amount of resources, money and time on “throwing a bone”.

Uh ohh .. it was NOT my intention to step on somebody's toe here. I am very sorry. 'tschuldigung.

Anyway - with the migration to Kotlin I have had developers asking themselves in which way they should continue to think. And - since some bugs (especially in regard to web-push to a dozing phone) haven't been taken care of in more than 15 months, I was just curious as to whether Google understands web-apps to be complementary to native apps or if they see web-apps to become as important and equal as/to native. Again - I don't want to sound disrespectful - I am genuinely curious in regard to Google's strategies.

Michaela

The misspelling of Referer.

In all honesty, though: It would probably be the removal of “synchronous <anything>”.

Pretty much all of the podcasting, tweeting, presenting is definitely part of my job although only useful to varying degrees career-wise (like shitposting on Twitter). And yes, care is definitely important when making public statements, as we do not run every tweet by our legal team (even when we tweet from @ChromiumDev. Our team is given a lot of trust, which makes our job so much easier.

For a while now we have been trying to embrace “Walk the walk, then talk the talk” as our motto, meaning that we should behave like real web developers. No cheating. Use the same tools, build apps from start to finish. Just that along the way we take notes about what is painful or difficult, where is DevTools lacking, what bugs did we find etc and feed that back to the appropriate people. And then we turn our lessons learned into (hopefully) useful articles, guidances and talks.

Squoosh is a product of that process and we are quite happy with the results and the reception :)

I don’t think TypeScript “takes over” JavaScript :D It’s synergetic, not adversarial. But that just as side note.

It’s hard to pick favorites, honestly. Wasm is definitely exciting and very fun. I am also looking forward to WHATWG streams being universally available. They are so versatile and flexible. WebGPU is also cool. SO MUCH IS HAPPENING.

No. Never. All my talks were absolutely flawless.

I have no hard data to support this, but I don’t think WebAssembly will help much. The problems we see are big monolithic chunks of JavaScript, forced layouts and forced paints.

JavaScript is extraordinarily fast considering it’s a weakly typed, originally interpreted language. In V8, JavaScript and WebAssembly have the same peak performance. Sometimes it’s harder to get there with JavaScript, though. So WebAssembly might make some cases faster, but the biggest problem is that there’s no easy way to break up work into smaller chunks and yielding to the browser in between these chunks.

Additionally, all of this is still happening on the main thread, when a lot of the work web apps do could be happening in a worker to keep rendering much more responsive.

Both, really.

Houdini, for me, is more of a “power-user” feature. I expect a couple of well-versed people to build modules for Houdini that everyone can then use. That doesn’t mean that the average web developer can’t write Houdini modules, they just might not get the same mileage out of it.

It’s practically impossible for one person to know every API browsers offer nowadays, so you need to know where your interest lies and what problems you encounter on a day-by-day basis.

Templates are Turing-complete.

For the record, C++ is not my least favorite language.

People who are better than me. I just love learning from my colleagues.

That being said, respect and ability to communicate are necessary for any form of collaboration. Can’t learn from someone who can’t communicate.

I am not a book learner. I remember reading Mastering Regular Expressions and the knowledge from that book is useful to this day. It’s also hard to recommend resources for “developers”. It really depends on what you are into and what you are working on. For example, if you are into graphics, Real Time Rendering is great to have, but if you are not into that, it’s gonna be boring!

Sausage Dog.

I think we need a proper scheduler on the web. Something like Grand Central Dispatch on iOS. With priorities, tasks and a way to move work off the main thread.

My colleague Shubhie Panicker has been working on a Scheduler API proposal. It’s just the first step and doesn’t do anything that I’d want it to, but it’s a great first step!

I recently upgraded to a completely over-the-top bonkers iMac Pro setup at work!

Most of the time you don’t need knowledge of data structures and algorithms. I think the important skill is to recognize that a certain problem you are encountering is, let’s say, a graph problem and that you can find and implement different algorithms to solve that problem. Often you can also just pull in a library that does these things for you.

That being said, having a deeper understand of what is happening will probably help you make more informed decisions with respect to performance.

Thanks a lot, Surma! 🌈🙌🏾😊

ahem

I think ECMAScript needs a way to move code between realms. Like this absolutely great Blöcks proposal.

My favorite feature would have to private fields 😉

Repeat after me: I will only animate opacity and transform.

Seriously though: The biggest problems I see are

• forced layout and paint
• bit monolithic chunks of JavaScript execution
• big images
• too much JavaScript

Run Lighthouse and ship less code!

I have been writing a lot of Rust recently and have thoroughly enjoyed it. It’s like the best parts of C/C++ and Haskell in one language.

Not sure which programming language I want to pick up next, to be quite honest. Maybe I should finally learn Python?