Twitter | Search | |
Mathias Bynens
I work on at Google and on ECMAScript through TC39. ♥ JavaScript, HTML, CSS, HTTP, performance, security, Bash, Unicode, i18n, macOS.
19,773
Tweets
779
Following
53,947
Followers
Tweets
Mathias Bynens retweeted
sMyle Mar 18
After over a year of work, a PR to update the ESM implementation in has been opened Some important bits * you can now use .js for esm by setting `"type": module` in `package.json` * node's specifier resolution is not on by default * .cjs 🎉
Reply Retweet Like
Mathias Bynens Mar 18
Replying to @Souvir
globalThis and dynamic important aren’t yet part of ECMAScript. In practice it doesn’t matter though, as they’re shipping in browsers!
Reply Retweet Like
Mathias Bynens Mar 18
Replying to @wesbos
ASI
Reply Retweet Like
Mathias Bynens Mar 16
I’d love to hear more as well. contenteditable seems fine as long as you don’t persist it to the server or to the URL (in which case you could be tricked into clicking a link that steals your admin cookie). Without cross-session persistence, folks could only XSS themselves.
Reply Retweet Like
Mathias Bynens Mar 16
Replying to @AmarachiAmaechi
1 MB = 1000 kB 1 MiB = 1024 kB Everyone using “MB” to refer to the latter is wrong.
Reply Retweet Like
Mathias Bynens retweeted
Daniel Ehrenberg Mar 16
Would it be useful for JavaScript and WebAssembly to have declarative feature tests, that can be run cheaply and early, and used to decide which code to load?
Reply Retweet Like
Mathias Bynens Mar 15
Reply Retweet Like
Mathias Bynens retweeted
Mathias Bynens Mar 13
📝 Should the web platform have first-class support for differential script loading? issue: More details: What do you think?
Reply Retweet Like
Mathias Bynens Mar 14
Replying to @jeffposnick
Bonus points if you make them `contenteditable`!
Reply Retweet Like
Mathias Bynens Mar 14
No takesies backsies, ! We shall henceforth be advocating for the moodle/noodle pattern.
Reply Retweet Like
Mathias Bynens Mar 14
Replying to @bmeurer
The proposal is now here:
Reply Retweet Like
Mathias Bynens Mar 14
Does it have to? Why would you need differential import() if the generated JS bundle that does the import()ing already knows which features it can rely on?
Reply Retweet Like
Mathias Bynens Mar 13
Replying to @jleedev @WHATWG
Not sure if joking... This proposal has nothing to do with ECMAScript version numbers.
Reply Retweet Like
Mathias Bynens Mar 13
This is 100% unrelated to any spec version number. I agree that would not be useful! Please read the proposal for details 👍
Reply Retweet Like
Mathias Bynens Mar 13
At any point in time, I imagine you’d ship maybe two module variants at most. Each of the feature sets essentially corresponds to a browserslist config.
Reply Retweet Like
Mathias Bynens Mar 13
Replying to @oriSomething @WHATWG
Please read the proposal. None of this refers to ECMAScript version numbers. I agree that wouldn’t be useful.
Reply Retweet Like
Mathias Bynens Mar 13
Replying to @_developit
It’s for the best. Holes are bad for performance
Reply Retweet Like
Mathias Bynens Mar 13
[citation needed]
Reply Retweet Like
Mathias Bynens Mar 13
Say it with me: extensions do not matter on the web
Reply Retweet Like
Mathias Bynens Mar 13
Indeed: you want a <script type="module"> fallback without introducing additional downloads in modern browsers.
Reply Retweet Like