IV

An update on Igloo

Changing priorities and more to come!

  • Written Icon

    Written:

    2026-02-24

  • Updated Icon

    Updated:

    2026-02-24

Project PageHomepageGitHub

TL;DR After a lot of battling myself between priorities, I’ve finally accepted defeat on making Igloo a Home Assistant competitor.

I would outwardly say I wanted to make Igloo a strong competitor, while making decisions that contradicted that. I kept making decisions because they were cool, interesting to code, or fast for the sake of being fast.

I took a big step back, many times, to really think about this. User’s won’t notice that Igloo has a better architecture, is faster, takes less resources, is more secure, or most of the things that makes Igloo interesting. I think ultimately Igloo would always be a platform that is a bit easier to use, a bit more reliable, and has a lot less device support.

But this has already been done! openHAB is a much better platform than Home Assistant. I hear almost only good things about it. And yet, people pick Home Assistant because it supports their devices and is just the automatic choice. Even if Igloo was better than openHAB, I don’t think it would become as popular.

What’s Next

So, if not a HASS competitor – what is Igloo?

It will mostly be my hobby project, used for my home. I will make decisions simply because I wanted to, or because I wanted to learn something, or it was interesting to code, or it was fast.

What’s Leaving

  1. Python support: Extensions are only Rust
  2. Dashboard System: No SolidJS, no module federation, no plugins. Just 1 Rust WASM file
  3. Backwards Compatibility: It will be pure anarchy
  4. Complex Query System: No matching strings, no glob patterns, nothing. Always by IDs or components
  5. Pi 3 Support: Dell PowerEdge R430 (my server) will now be the primary target. It may work on ARM, but it’s not a goal. The platform is meant to scale on many cores.
  6. Intuitiveness: There will be no priority towards being intuitive or to be used by non-programmers.
  7. Reliability: The new program has a lesser focus on reliability. This means we now allow unsafe!!
  8. Package Manager: Packages are manually installed.

What we Gain

  1. Ridiculous Performance: No more JSON or Bincode over Unix sockets. We want zero-copy, less context switching, and fully shared device tree (for parallel query execution).
  2. Faster Dashboard: No JS, no JSON, no plugin loading.
  3. Better Ergonomics: Only targeting Rust extensions means no code generation, no designing systems to be cross-language, nothing. Everything is Rust and we leverage that.