Joe Fabisevich

The ChatGPT model picker showing a long list of models

We're less than 24 hours away from this list getting even bigger. 🥲

// The SavelLinkButtonTooltipView is positioned by deriving the global position of the save button. // The save button can be positioned on either end of the screen, // so we need to control where the arrow icon above it is positioned. // // How does this system work? Like a well-constructed but insane trampoline.

Found some great docs that I'd left for myself in Plinky.

Been thinking about the transition of iOS from skeumorphism to flat design to liquid glass. I think my biggest takeaway is that adding personality to your app initially was all graphic design, then became visual design, and now is shifting to interaction design. No strong feelings yet, just vibing.

Your career trajectory has been tremendous, your 300th episode sounded unrecognizable from your first.

Plinky – Easily Save Links | Enjoy Them Later Plinky makes saving links effortless. With one-tap saving from any app, browser, or integration, you can organize articles, videos, and more with folders, tags, and reminders. Find links instantly wit...

Orrrr you can try @plinky.app to save links from Bluesky *and* any other website or app, with Tags, Folders, Search, Reminders, Pins, for a fully curated, organized, and customizable link-saving experience. :)

Break Liquid Glass in case of emergency.

The grandmaster diet: How to lose weight while barely moving You wouldn't think that chess players, who sit for hours on end and extend their arms only from time to time, would struggle with weight loss. But they do. Inside the very real, very bizarre metabolic...

Not exactly scientifically proven, but that reminds me of this.

Build.ms by @mergesort open graph image

It’s not a secret, new open graph image for my website. 😄

A bunch of text that says something like AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

I did it, I finally drove Claude Code insane enough to start yelling at me in computer.

A bunch of text that says something like AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

I did it, I finally drove Claude Code insane enough to start yelling at me in computer.

Ah yes, you must mean notable data philosopher Derek Parfait.

My suggestion would be if you can afford $20, just try Pro. Worst case it doesn’t work, best case it does, the most likely outcome is that you get a feel for the tool and can make a more informed decision.

Max. It’s not cheap, but as an indie I would easily pay $100/month to be able to check things off my roadmap faster, so I “happily” pay for it.

I use it for both. Probably 90-95% of new code I write is AI generated — but of course it's extensively planned and code-reviewed before ever going to production. The results are fantastic now that I've developed the relevant skills, and the workflow is faster since I can work on two things at once.

Absolutely, it’s heads and tails above any competitor. A statement like that may only be valid in the AI space for a few months, but Anthropic has absolutely built the best agentic coding tool in the space. I also use it from all sorts of things, but it’s most valuable for coding.

Most importantly, scrolling is about to get sooooooo much smoother, especially for people who’ve saved thousands of links.

Spent five days working with Claude Code to optimize Plinky's use of Boutique during large link syncs, and I'm not exaggerating when I say the speed and number of Store writes that bottlenecked the system has dropped thousands of percent. And now I have 1,500 more lines of tests to keep it that way.

Don't boo, use Boutique.

Maybe, kind of?
1. No subscription offering will get you to Twitter scale.
2. The only reason those open source projects exist is because of companies like Twitter, Google, Meta, etc.
3. But if you somehow reach Twitter scale with them, who invents the next thing to continue your growth trajectory?

Twitter invented many, many technologies to achieve the scale it did, you can’t just scale vertically. But even if you say “just buy more servers”, by the time I left Twitter was managing nearly 750,000 servers in its data centers, which required its own innovations to maintain and manage.

Also leaving your S3 bucket unprotected on the public internet, that’s a real failure.

I guess history doesn’t repeat, it echoes.

Much better results today. 😌

That makes sense to me, I’m just curious because there seem to be use cases that don’t seem like a natural fit, but could be massaged intro working. I still don’t think it’s by any means guaranteed to be the right solution, and I’m mostly compiling examples of where it falls short for a talk.

Sounds like a good reason! I’m documenting all of these, so can I also ask for more specifics?

1. What kind of app is it?
2. What kind of feature were you trying to build?
3. Did you try approaches to work around the problems you ran into, or did you intuit that it wouldn’t work for your use case?

I think you’re thinking of Rust. 🦀

atharva @AtharvaXDevs Follow last night, I was skimming TanStack Query docs and went down a rabbit hole on its server-side usage with Next.js. What I found is a total gamechanger for performance. @tan_stack > it's a pattern that combines server-side prefetching with client-side hydration to practically eliminate initial page load spinners. When you pair it with Server Actions for mutations, your app feels insanely fast. here's how it works, a Thread TanStack Query v5

Every few years someone in the JavaScript ecosystem reinvents PHP and thinks they've stumbled upon something brilliant. You've created markup that signals to the server declaratively how to fetch data, congrats! Wait until you read a little more to learn about the potential downsides too.

For people trying out FoundationModels: What have been your biggest issues so far? Specifically problems that would prevent you from considering the framework for production use.

I appreciate the advice but 1,000 passions sounds like too many.

Oh it happens, but thanks. 😄