Jake Lazaroff

"why do you use safari?"

Blink: Intent to Prototype: HTML toolbar element

would love for this lil web component i wrote to become obsolete!

npmx.dev/package/rovi...

I'm so tired of automated thinking machines.
Make my thinking *better* or it's worthless.

AI gets better if it makes humans better

i can't get over describing the coalition of the mayor of new york city as "young people in cities" 😂

WebHaptics – Haptic feedback for the mobile web. Haptic feedback for the mobile web.

this is so cool
haptics.lochie.me

maybe find a way to cope without telling jews to accept the support of someone who wants to murder us

if you are telling me to welcome the support of a neo-nazi i'm gonna assume that you're a neo-nazi as well

modern CSS is so good

Another custom select list using using appearance: base-select.

On browsers that don't yet support this, a normal select list will be displayed.

See the full demo on @codepen.io codepen.io/cbolson/pen/...

Learn Yjs by Jamsocket Learn Yjs is an interactive tutorial series on building realtime collaborative applications using the Yjs CRDT library. Learn about handling state in distributed applications using Yjs shared types, w...

thank you so much! i actually also built learn.yjs.dev 😁 (with @paulbutler.org and @taylorbaldwin.bsky.social)

i mean, i hope you're right! but i don't think that's a forgone conclusion

i agree but you don't also see LLMs as a threat to the web, in the sense that people may come to use them as an intermediary to summarize websites without visiting them? there are a lot of techbros who want their chatbot to be *the* information gateway and it's not implausible that one will succeed.

i think if you aren't sure what the semantics are then it's fair to say that this is not actually standard functionality that everyone already uses

i do too, but the level of effort required today to write a full-blown HTML parser with declarative bindings is orders of magnitude higher than: given that templating system, write a few lines to add syntax sugar for event listeners

same question as with the @event thing though — why does this need to be part of a native templating API? you could do this in userland with whatever semantics you wanted by writing a tagged template function.

sure. the line is subjective. i think the crux of our disagreement is that i see a native templating API as a low-level tool on top of which to build abstractions, rather than something app devs should be using directly. i'd cover the slow/error-prone parts rather than add sugar

just my two cents 🤷‍♂️

it just makes the web platform as a whole more complicated than it needs to be. like, browsers now have to support popovertarget and popovertargetaction in perpetuity even though command and commandfor are an all-around better API. all these little papercuts add up!

right but what does this mean with regard to HTML templating? like, what are the semantics? does a function get `children` as a DocumentFragment? does it need to return an Element? can you pass other "props" to it? etc

i think they are

that's my point — with directives/custom attributes/etc it's easy to write your declarative event binding in userland. if we standardize @event and .prop… i mean, it doesn't *totally* close off the ability to use that syntax for something else in userland but it would make it extremely weird

unclear to me what html`<${Fn} />` would do?

i meant your example. <button ${event("click", () => console.log("hello"))}> builds it into the template system in a more general way.

imo the platform is just the wrong place for syntax sugar like this. easy to write a tagged template function that converts @click={foo} to ${event("click", foo)}

is that not building it in? the issue with the lit syntax isn't whether it's possible, it's whether to add a very specific syntactic sugar on top of it.

a DOM templating API would let you pass arbitrary javascript objects, right? so you could in theory do something like

<button on-event=${["click", () => console.log("hello")]}>

frankly the syntax is the least of my concerns. i don't think the functionality should exist at all. i'd much rather see
something like custom attributes, which would let us implement the same thing much more flexibly in userland

i mean fwiw the .prop and @event prefixes feel pretty idiosyncratic to lit

yeah trying to get htm playing nicely with typescript led me to that issue too. i think it would be a big squandered opportunity to add something peculiar to html templates rather than a primitive that we could use for generalized parsing of template strings.

idk if i'd call it visual noise… the whole reason i'm not using raw hyperscript in the first place is because i find HTML-ish syntax easier to read.

i think the function call syntax is tolerable if you use it for a few leaf components. if you have a bunch of complex or nested components it's rough.

both approaches kinda suck though imo!