dave.js

Someone needs to rebuild Arc. How hard could it be to fork Zen and replace Firefox with Chromium? 🤔

Seeing a lot of hype/controversy around TOON. I’ve noticed a strong correlation between those dismissing TOON and having #opentowork on their LinkedIn. Hot take: It’s too early to evaluate the long-term utility of TOON. Those not giving it a chance now are just cosplaying as pragmatic engineers.

LLMs have ruined my favourite punctuation mark — the em dash.

Cursor's Composer model really flies in agent mode and the results are pretty accurate.

This feels like a critical shift in my dev workflow. I'm now reaching for agents at least twice as often, and it's always Composer.

#ai #vibecoding

@cursor.com.web.brid.gy

That would allow for platform implementations to diverge though and experiment before standardizing common functionality. This is starting to feel like a web standards proposal now 🤔

…Namespaces could help here if we failed builds when unimplemented directives are found. Some directives might get standardized and moved to the global namespaces.

This reminds me of vendor prefixes in CSS though, so I’m a bit hesitant.

In a monorepo, multiple frontends could be deployed on different platforms. In this case the lines get blurry and it’s unclear what code is and isn’t portable between projects. The code deploys fine, but doesn’t behave the same…

Yeah, namespaces could help with that. Like “vercel: use workflow”. Tanner mentioned this too, but that doesn’t fully solve the problem because it’s still up to everyone to cooperate. Better than just all global though.

Versioning is a great point! Haven’t even considered that yet

I think we do need to solve for some of the issues Tanner raised though. Like having an ergonomic syntax for options and being able to differentiate between language and platform-/framework-specific annotations. If we can solve that with directives, great, but I haven’t seen any good proposals yet

That’s a fair point. Counter argument though: we already have restrictions around placement of hooks even though they’re just regular functions and we use lint rules to enforce placement. We could do the same here.

What concerns me most about directives is it’s not clear which ones are a language feature vs a platform/framework feature, and if you move your code between platforms that both have “use workflow” but diverging implementations, then things can break unexpectedly even though the code is identical.

Hmm, since decoratiors aren’t official, what about tagged template literals?

vercel`workflow(options)`;
function sendEmail() { … }

I don't think there's a single release from GitHub Universe that doesn't involve AI 🤖

Shoving AI into every feature seems like a waste of compute for the value gained.

Kaelig giving the announcement on stage

I’m excited to announce the first stable version of the #designtokens spec is live! v2025.10 was just announced at #schema. Learn more at designtokens.org.

Yann Braga | Storybook Vitest | ViteConf 2025 YouTube video by ViteConf

Storybook v10 is looking amazing 🤩

The new CSF, the deeper Vitest integration, so many DX wins!

@storybook.js.org @vitest.dev @yannbf.bsky.social

youtu.be/YNQ_cJ6yy2Q?...

decorators-as-alternative-to-directives.md GitHub Gist: instantly share code, notes, and snippets.

I haven't seen Decorators discussed as an alternative to directives so I've drafted some thoughts here: gist.github.com/davejsdev/1e...

Might write up a more detailed blog post about it unless this is just a bad idea.

cc: @tannerlinsley.com @t3.gg @tanstack.com

a man in a plaid shirt is standing in a living room and says what have they ALT: a man in a plaid shirt is standing in a living room and says what have they

I use Tailwind on nearly every project now. Not because I like it, but because LLMs like it.

Hot take: Modals suck. What's the point of partially obscuring content? Just make a choice: either expand the modal content inline OR display it in a standalone view.

#ux #design

So... are we using Biome or oxlint/oxformat now?

I envy your result. The lower the score, the more intact your sanity is 🤪

new Date("wtf") How well do you know JavaScript's Date class?

I scored 18/28 on jsdate.wtf and all I got was this lousy text to share on social media.

AI Code is Obsolete The feedback loop that could stall software evolution.

Reflecting on my experience with using AI agents for scaffolding projects with @bolt.new! 🤖🧑‍💻

#ai #agenticcode #typescript

open.substack.com/pub/davejsde...

Or would it work with @convex.dev (and other sync engine tools) for more efficient syncing of local data?

Exploring the TanStack DB docs, @tanstack.com is this an alternative to @convex.dev? 🤔

Example: my mother worked in healthcare doing home visits as recently as 5 years ago (since retired). She shared a desktop computer with 3 other coworkers, used paper charts and fax machines, and was given a flip phone for comms while out on visits. Many people still work like this in 2025!

These problems are everywhere, but as tech professionals, we’re often too isolated from other professions that it’s tough to identify them.

For about two decades, if your idea didn’t involve “disrupting” an entire market, it wasn’t worth investment. This has created a huge gap in the market where we still have a lot of unsolved problems simply because there’s a limited (but still profitable) market.

There’s also another problem though trying to emulate these successes. For a long time, venture capital has been focused on finding the next unicorn, which has led to a lot of money being pumped into moonshot ideas to make $10B+ and dismissing ideas that might top out at a few million.

This is such a great opportunity though, and the unicorns we all know are typically startups who have taken a confident step outside of the tech echo chamber and solved problems outside of tech: Uber, AirBnB, DoorDash, Spotify, Zillow, to name a few.