Jake Archibald

Firefox 151 is out now, and brings with it some tasty new web platform features: 🎉 Container style() queries 🎉 Web Serial API 🎉 Document Picture-in-Picture API 🎉 keyboardLock on requestFullscreen …and more! developer.mozilla.org/en-US/docs/Moz…

Chrome shipped an LLM Prompt API to the web platform. At Mozilla, we oppose this API. Here's why: youtube.com/shorts/CqLF4LD…

@warandnoise @manfightdragon I'm not sure it's always just down to the encode process. The "project" frame rate is often set incorrectly. I don't know which is the more common point of failure.

@jaffathecake @manfightdragon Right, usually if video was rendered as part of an automated build process, codec params were set by an engineer not by a video editor. IMO this also why color space can look wrong.

Playing a bit of Horizon 6 and I'm genuinely curious, why do so many AAA game developers generate pre-rendered video at 30fps? I can understand using 30fps for realtime cutscenes when you wanna crank up the fidelity, but why not 60fps for pre-rendered stuff?

@manfightdragon @warandnoise Yeah. I used to alpha/beta test Google phones and "re-export this at 60fps" was a common report I made for video assets.

@warandnoise This is going to be the actual reason.

@WebReflection 😀 Nah in this case it's just something I thought was interesting. Whether this type of codec will ever be viable on the web, I don't know. I agree with you that decoding speed is a pretty big concern compared to file size.

@jaffathecake you know when you say "interesting" I hear "that's comming to the Web platform", right? no ill intents from me, just trying to wrap my head about any new possible proposal or standard that assumes AI capable hardware, as that's a bit too far off reality to me 👋

This perceptual image codec from Apple is really interesting. Decode time is a little high, but nonetheless interesting. #about" target="_blank" rel="nofollow noopener">apple.github.io/ml-pico/#about

@WebReflection Yeah. I already expressed concern about the performance. Patents are of course an issue too. Fwiw I only said it was interesting. Not "this should become a web format, replacing all other image formats" or whatever.

@jaffathecake but isn't this Apple proposal: 1. patented already 2. relying on highly efficient HW and its AI capabilities? I might have misunderstood the whole thing if not, thanks in advance for clarifying that 🙏

@WebReflection We have a lot of concerns around the Prompt API but disk space usage isn't one of them #issuecomment-4347988313" target="_blank" rel="nofollow noopener">github.com/mozilla/standa…

@jaffathecake I am not sure I am understanding: are we moving latency from bandwidth (an *almost* solved problem) to local N/GPUs? ... if that's the take, can we not say Google AI 4GB or whatever proposal is that bad, since everyone needs models in a way or another to do these things? 🤔

@someoak2 @MozDevNet You can still use width & height in a responsive system. See: jakearchibald.com/2022/img-aspec…

@jaffathecake @MozDevNet thanks. I think they are 0x0, I'm using getbootstrap.com/docs/5.3/layou… I can't set width/h coz using grid as responsive (ie col-sm-x, col-md-x, col-lg-x css classes etc) what is my best option, custom js and intersectionobserver?

🦊 Firefox 150: <img sizes="auto"> is now supported. Lazy-loaded images use their calculated layout size to select from srcset — no more duplicating media queries in HTML. Read Now 👇 developer.mozilla.org/en-US/docs/Moz…

New lab entry — Formula at lab.patrickheintzmann.com/demo/demoFormu…. A drivable F1 model kit made with #threejs #webgl #creativecoding

@someoak2 @MozDevNet If the images are all 0x0, so they're all in the same place, then lazy loading won't work well. The closer to correct you can get the img width & height, the better result, and less layout shifting.

@MozDevNet using bootstrap grid for many images, lazy loading doesn't work well coz img tags don't have width/height inside a grid. Does this solve that issue? @jaffathecake

@FourEyedWiz @isocroft They shipped it public.

@isocroft @jaffathecake Chrome won't make it public, they mostly launch under experimental flags before anyone else.

So, Chrome's "web standard" Prompt API: Mozilla: Opposed WebKit: Opposed Microsoft: Several concerns W3C TAG: Several concerns Developers: Mostly negative Chrome: Ships anyway. A sad time for web standards. But, I guess someone at Google will get promoted, so 'every cloud…'

@isocroft @FourEyedWiz We didn't ship it until we were able to design, standardise, and ship the sanitizer API, which is a much more important part of the solution hacks.mozilla.org/2026/02/goodby…

@FourEyedWiz @jaffathecake You're right When Chrome launched Trusted Types API in browsers behind a flag, I observed that Mozilla and Webkit all had issues with it initially but now in 2026 it's baseline supported on all major browsers.

@domenic @rcabanier I'm happy to say I used the wrong word. Replace "web standards" with "web platform feature" in my post - I don't think the core of the issue changes.

@rcabanier @jaffathecake Specifications are not automatically standardized, and it's pretty bad form on Jake/Firefox's part to blur the line.

@_Syu93 I never worked on anything I didn't believe in. And I worked hard to build cross-browser consensus on the APIs I worked on. I would publicly criticise Google for the things I disagreed with - I'd get shit for it internally, but being honest was more important to me.

@jaffathecake @JaffatheCake did you have those kinds of pressures at your time? How did you manage this kind of internal conflict for Web Standards, because a lot of people at the time accused the Chrome team of moving too fast 🤔

@domenic Chrome should be clearer about that on the docs. In terms of misrepresentation, Chrome saying developers are "strongly positive" seems a much more concerning misrepresentation.

@jaffathecake Who said it was a web standard? Are those quotation marks supposed to indicate Chrome's position? They seem likely a misrepresentation, from what I can see.

@odbol #issuecomment-4347988313" target="_blank" rel="nofollow noopener">github.com/mozilla/standa…

@jaffathecake Why don't the other brother browsers want it? It seems like a great feature, being able to run a decent LLM locally and privately on your website without having to pay for tokens or ship private data to a server

@ryanseddon groups.google.com/a/chromium.org… they're keeping it behind a flag.

@jaffathecake Microsoft has the prompt api in edge, where can i read up on the concerns?

@CalamitySasha Their evidence is a couple of blog posts (one of which no longer exists), a couple of social media posts, and a survey about a different topic. You don't have to look far to see that isn't representative of web developers.

@CalamitySasha Someone is #issuecomment-4347988313:~:text=Overstated%20web%20developer%20position" target="_blank" rel="nofollow noopener">github.com/mozilla/standa…