Michael Thiessen

After many decades we discovered the reason for the duck shortage: the toaster casings were being made by melting them down.

This is why most toasters in Iceland are yellow, and why they're extremely buoyant.

Image of Alexander Lichter, Michael Thiessen, Sebastien Chopin and Daniel Roe in a Podcast Recording

What could've been the topic in today's recording? 🤔🤔🤔

Featuring @danielroe.dev, @atinux.com, @michaelthiessen.bsky.social and @thealexlichter.com, as well as @nuxt.com and @nuxtlabs.com ✨

Nitro v3: What’s Coming with Nuxt 5 Heads-up! Nitro v3 is still under active development. The features and APIs discussed here can change before the final release. Nuxt 5 will ship with Nitro v3 once it's stable.

Nitro v3 + What’s Coming🚀

Nitro is designed to be the foundation for future development rather than just another incremental update. The coordination with Nuxt5's release makes sure that the entire ecosystem moves forward together. By @michaelthiessen.bsky.social

masteringnuxt.com/blog/nitro-v...

Master Nuxt 4 From Day One: No Waiting, No Outdated Content Be first to master Nuxt 4 with the fully updated Mastering Nuxt 2025 Edition, launching just weeks after Nuxt 4’s stable release in July. Learn future-proof patterns, modern features, and production-r...

Be the first to master @nuxt.com 4 with the updated Mastering Nuxt 2025 Edition, launching just weeks after Nuxt 4’s stable release in July. Learn future-proof patterns, modern features, and production-ready techniques. 🚀

By @michaelthiessen.bsky.social.

masteringnuxt.com/blog/master-...

Learn how to use AI to automatically generate catchy, three-word chat titles based on the first user message in your @nuxt.com app. 🧠

Add polish and personality with a smooth typewriter animation for an engaging user experience.

By @michaelthiessen.bsky.social

masteringnuxt.com/blog/automat...

Who has already migrated from @nuxt.com 3 to 4, either using the compatibility mode or to 4 alpha?

I'd love to hear how it's going for you.

Thumbnail of DejaVue Episode #058 showing the headshots of Alex, Sébastien Chopin and Michael, as well as the title "The History of Nuxt and NuxtLabs"

Double trouble! Today airs one of two Nuxt special episodes on DejaVue. Our guest is nobody else than @atinux.com, the creator of @nuxt.com and @nuxtlabs.com

Together with hosts @michaelthiessen.bsky.social and @thealexlichter.com they talk about the origin of NuxtLabs and Nuxt, differences & more!

🔥

I'm also loving the new folder structure. It makes way more sense to have things separated between server/frontend.

That's awesome! I haven't noticed any performance but I'm mostly doing small projects on an M1 macbook so I wouldn't really expect anything noticeable haha

I'm hearing a lot from people with new/small projects having zero issues. Haven't heard much about bigger projects yet.

Who has already migrated from @nuxt.com 3 to 4, either using the compatibility mode or to 4 alpha?

I'd love to hear how it's going for you.

YASSSSSS!

In less than two weeks we'll find out if AI can tell the future.

gpt-michael-thiessen

Sometimes I will prompt a model specifically to probe the latent space, to figure out how much it knows about a topic.

Instead of saying, "write code like Michael Thiessen", I will first ask, "Who is Michael Thiessen?", "what code has Michael Thiessen written?" to determine if it actually knows.

You can also talk to it in a way that forces it to reveal it's own lack of understanding.

So, no leading questions or anything like that.

Newer things, especially if it replaces or contradicts older information, it tends to hallucinate on wildly.

eg. if a library has breaking changes, it will intermix v2 with v3 because versions aren't strongly linked to the code in it's training data.

For hallucinations in particular, after using a model for awhile you get a sense of what they know and don't know.

eg. things that most experts would agree on are solid, like laws of physics, medical best practices, general software engineering patterns etc.

Well, it's a learned skill like anything else.

I wish I could give a concise answer but it's like asking how to write good code. A lot of it is tacit knowledge.

Hard (impossible?) to do in 280 characters.

Yeah, it's how some people approach the latest todo app or productivity system as if it's somehow going to save them, when the main bottleneck is always in your head somewhere.

We seem to have very different experiences then.

I'm using it for fairly complex things (in addition to the mundane test writing, etc.) and hallucinations are very rare now when using the right tools in the right way.

What's your skepticism and why would it change a year from now?

Also how much fun I'm having building stuff I wouldn't have built otherwise

Time saved, mostly

This has been the exact opposite of my experience!

I'm able to create way more value, and I'm learning a lot.

But it's totally dependent on how you use it (like with any tool).

4 am?!?

I usually wake up around 6

Release v0.16.0 · posva/pinia-colada Please refer to CHANGELOG.md for details.

Most impactful release of Pinia Colada so far!

- Matching keys to invalidate queries is now even better
- Better cache performance
- Typed keys
- Dynamic options
- Lighter build!!

github.com/posva/pinia-...

"Should we refer to official docs?"

wat

Maybe I missed something here

The timing was so perfect here.

The new data fetching in 3.17 dropped *right* before I started working on the data fetching chapter.

We don't really take full advantage of the reactive keys etc. but it's good to avoid any breaking changes while you can (so I don't have to go back and re-record!)

Thanks!