Are there any solo adventures?

This is exactly my dream! I hope it happens!

I’ll check out that module, thank you for the tip.

I feel like the Codex should give them much of the game engine already and I’d love to play an adventure written by the team. My busy adult friends bounced off Draw Steel! for the time complexity

My other favorite would be a “Draw Steel!” game by @hellomcdm.bsky.social in the style of Final Fantasy Tactics / Triangle Strategy

I would love to see a Blades in the Dark game as well!

Yeah, totally agreed

What did you change about the system prompt?

That’s a really cool idea for the map

I assume most people view source through dev tools and it’s pretty easier to unminify with formatters.

It seems like the evasion / accuracy die is working a bit like armor with a dynamic value?

I’m very curious about this one

Our reactivity was lazy too. In practice that didn’t matter when you rendered a large list of reactive tasks.

Are there any times where you feel like you know more than Opus? Eg are you providing value add by steering the model?

That makes sense. I was thinking about building a local first app so I don't even need SSR. I'd use TanStack Query to just load the data from IndexedDB

@jovidecroock.com do you have a perspective on the ideal Preact stack these days? I imagine some combo of ISO Router, Tan Stack Query, Signals?

That’s really cool! I had not noticed that in the API before. I had been doing a similar thing with classes by making the signals private properties and exposing getters / setters for them.

Yeah, I think this was one of the core issues with how Luna 1 worked. Everything was a fine grained leaf including responses from the server and all component state so the graph was very large.

Totally agreed. Do you have a sense of the practical size limits for Preact Signal's graph? E.g. are there best practices of when store an object as a signal vs an object of signals? Same for collections?

I really enjoyed programming in it and want Signals to work. (I use Preact Signals in side projects, it’s great!) I just worry about repeating the patterns that drove us to rewrite Asana.

So we would setup and tear down large dependency graphs that would never be leveraged by the user because of their behavior. Also many of our engineers struggled thinking about frequently their computed out effect function would be triggered by dependent inputs.

We did end up moving away from fine grained reactivity due to performance though. We built our entire framework around Signals (it looked a bit like Solid and was the inspiration for Meteor). In practice, we didn’t get the fine grained reactivity benefits. Users would load a page and navigate away.

Yeah, I’m not saying the model is entirely broken. Just that users need to know about those edge cases. We wrapped time and other mutable globals in Signals to help. Kinda like the useSyncExternalStore of Signals.

Asana built a whole framework on Signals in 2009 that we eventually moved on from. There are a few other issues that can cause bugs. If you use a non signal value in a conditional that can cause issues as well (e.g. time or randomness). These issues also apply to computed signals too right?

Can you make it so that the bottom of the subtitle aligns with the tip of the sword?

What’s the subtitle?

I like the sword of the pixel logo but the glow and smoothness of the word

Cool! That type of project has always been my dream side project.

@rpdeshaies.farirpgs.com what is Free Blade?

I’m so excited! Given that you don’t want to build transitions / form actions, do you see Preact diverging more from React?

Yeah, I’m curious as well.