That sounds like it'll be easier to maintain. Will it also be faster?

It is currently faster than the stock Candle / MPS shaders it replaces on MacOS/ARM64, and IIRC a bit slower than OpenVINO/CPU on my old Windows laptop, where I never got OpenVINO/GPU to compute correctly. Candle didn't have support for GPUs on MacOS/Intel, and OpenVINO ceased to be supported there.

Compared to OpenVINO (I tried ONNX runtime too, but never got it produce correct outputs with my quantized models) it is very nice to be able to build the exact kernels I need, at the quantization settings and precision that works for the models I have and with the custom operators required (speech models do a lot of non-standard stuff), run from a single set of sources, and not have to ship a hefty third-party DLL, and having to deal with their memory leaks and other stability issues.

They should adopt Nushell! It would be seen as very "Apple"- a brave great somewhat non interoperable user experience upgrade

I was a big fan of Nushell for some time but has the same problems as unix shells. I have this CLI tool that manages a lot of things for me (a big one is downloading and cataloging arxiv papers, it downloads them and tags them with metadata and gives each PDF a proper filename based on the author's name and publication year). I tried building this in Nushell but it became eventually so painful.

I'm sorry, what are the same problems you're referring to? I ended up leaving because:

- it was unstable and kept changing (less of an issue now)

- I still need to know bash due to its ubiquity

- it was worse than python for data crunching scripts (ide tooling + my familiarity)

This was several years ago so I'm curious about a more recent perspective

There is some unstability yes.

For me the larger problems were the lack of types which in practice meant that I was running into the same problems as with bash.

Nushell does have types: https://www.nushell.sh/book/types_of_data.html

Support for custom types is not there.

Oh you mean algebraic types (Elm terminology)? Or the newtype pattern? Or something else?

I'm assuming this was a while back, mind sharing how far he went or what he was able to produce with this focus?

He was planning to go into academia, but ended up abandoning his PhD ABD because he got exposed to the negative interpersonal politics that can happen in academia and realized that was not for him.

So he went to go work a dozen years at a hedge fund and then retire to be a stay at home dad.

His primary research interest was logical frameworks, constructive mathematics, etc. His PhD topic was on using linear logic to verify concurrent programs.

Good for him! That's the dream!

CRDTs are eventually consistent data structures

Interesting, but I'm not sure how it's different from Observable Framework ( https://observablehq.github.io/framework/ ) for dashboards as code. I like that one because it's Markdown and JS instead of a language defined in YAML + templates

The use case that immediately came to mind for me was to use dbt charts as a rendering method for an MCP App.

You don't really want to give a LLM client code execution environment like Observable Framework does. dbt charts appear to validate the entire yaml input including the SQL being sent to the data source.

So I'm getting: a data vis layer that can be dynamically defined and rendered via LLM/MCP client without the headache of sandboxing a JS Runtime environment.

I like the observable framework too - very flexible, you can nail very specific visualization details by dropping down to JS/html.

Maybe this is targeted to pure SQL folks that can get lost in UI details.

PS: one issue that I have with observable framework is authentication and authorization , it requires some system to be built on top of it to handle authn.

Maybe it's because I'm not an RBAC expert, but I had a hard time understanding the data model for folders. Are they a subject? And then there's a layered CASL where layers overwrite each other? And the user is an actor?

Pop a screenshot in your README please!

I just started diagramming with d2- I'll have to try this one out!

The great thing about using a package manager is that it also handles updating and uninstalling software, not just the initial install

Sure, but the point is (1) that you don’t need a package manager to install software, which the comment I replied to seemed to assume, and (2) that for something you’re developing against like SQLite, installing it via package manager really isn’t that important. It doesn’t need to integrate with the rest of your system the way typical apps might.

What seems to be happening here is that people have learned that package managers are the right way to install software, but they don’t really understand the reasons, or where and how exceptions might apply. IMO your comment does that as well.

Even for a personal project such as the one in the blog post, I'd still rather use a package manager to install SQLite.

It's POSSIBLE to build it from source, and manually re-download the source and C compiler (and any dependencies, though I'm not sure SQLite has any) when I want to update and instruct any collaborators to do the same thing... But I'd still rather automate it all with a package manager.

Especially because I'm probably also handling other dependencies in the project and I value the uniform simplicity of using a package manager for all of them

How are you using dbos with SQLite? The docs say its a Postgres library

Hello, I'm from DBOS.

Our Golang and Python SDKs support SQLite as well as Postgres.

Interesting, it's not mentioned anywhere in the https://github.com/dbos-inc/dbos-transact-golang README. By contrast, Postgres is mentioned right at the top.

I recommend updating the README to mention SQLite somewhere, with maybe a link to https://docs.dbos.dev/golang/reference/configuration#using-s... ?