Michael Muré

@michaelmure.bsky.social

I just realized ... I want to use normal git signing keys so that git-bug naturally mesh with a normal dev worflow. But I can also publish those keys in the did:plc registry to have a centralized (vs one copy per repo) identity handling, with a key log, local key signing and everything!

I wish Github would expose each user's log of public key, not just the currently valid ones. Then I could anchor signed operations of a CRDT (git-bug) on those. DID:PLC does expose that, but what's the story for locally managed keys these days?

I wonder if having LLM personal assistants (safe bet at this point) will push businesses to expose standardized APIs: book an appointment or restaurant, product search ... Hopefully those don't get centralized into the same usual giants ...

I'd like to announce github.com/MichaelMure/..., a go native @automerge.org implementation. It doesn't have all the bells and whistles of the rust version (no sync protocol/repos), but has the core CRDT logic and should be reasonably fast. It produces byte identical data, and has the same license.

GitHub - MichaelMure/gotomerge: A partial implementation of automerge in go

A partial implementation of automerge in go. Contribute to MichaelMure/gotomerge development by creating an account on GitHub.

github.com

I feel like there is a gradient of software engineer going from very controlling of the code structure, to more or less appending features however it comes up. Coding assistants are going to magnify this gradient like never before.

One major distinction between software and woodworking: when you are tired/frustrated, it's way more important to stop for one of those. A computer can't take your hands but those machines definitely can.

I'm opening a new personal tech tree! Today, some CAD for a shoji (japanese sliding door). It's all parametric so I can adjust the global parameters (dimensions, kumiko counts, ...) and all the cuts, mortises, tenons are recomputed. Now the questions will be: can I make it for real?

exploded view of the tenons/mortises of a shojifull view of a shoji in CAD

I wrote maybe half of the decoding of an #automerge document in go, as an exercice in performance and database-like internals (column storage, RLE..). It's sitting in a private repo until I come back to it, but happy to share with a foss license if there is interest. Just ping me.

Recursive hierarchical explainer for LLMs: starts with a high level concept ("explain database") and get a breakdown. From there you get an UX to expand as deep as you like on sub-topics ("LSM trees") in various ways: more explainers, implementation, connective question... Does that exists?

Random thought: full text search engine use word transformation, canonicalisation to map the input text to some internal index (my understanding). Could a LLM vector space be used instead? That would remove the need for the transformations, and make it work cross-language.

> Excited to try this. > This is incredibly cool. I love seeing local first software starting to make a comeback. > I've been yelling 'omg why doesn't someone build [...] for YEARS. This is wildly exciting. More than coffee, those are what feed an open source project.