Matt Teichman

Cool; that was my impression. This project really seems to open up some possibilities for AI engineering that I haven't seen before; looking forward to following further developments.

@trygrace.dev I was just playing around with the Grace repl, built from source off of GitHub, and was wondering whether it was possible to hook it up to open weight models run under ollama (or similar) to generate typed data.

@mikemull.bsky.social I was coincidentally just playing with crdt mode in Emacs earlier this week. Works shockingly well.

Our Flag Means Local-First - Picture Me Coding This week Mike and Erik talk about the local-first software movement. There's a pretty cool paper about it from 2019 called "Local-First Software:You Own Your Data, in spite of the Cloud", and there's...

Absolute banger of a @picturemecoding.bsky.social episode on 'local first':
www.picturemecoding.com/2222783/epis...

!!!!!!

ooh yeah, would love to learn more about this take

Fun fact, this is legal Haskell:

>>> (print <> print) True
True
True

getting ready to move in here
photoarchive.lib.uchicago.edu/db.xqy?one=a...

\{ key : Key } -> { "You'll usually use keys for API calls": prompt{ key, text: "hi" } , "The key can never rendered by the client": key } --- You'll usually use keys for API calls: Hello! How can I help you today? The key can never rendered by the client: 🔒

The Grace programming language now natively supports a Key type for managing API credentials. The benefit of a distinct type is:

- Grace guarantees that values of this type are never rendered
- The Grace browser obscures form inputs of this type

like in swordfish

damn straight

I'm pleased to announce OxCaml!

OxCaml is Jane Street's branch of OCaml. We've given it a new name and a snazzy logo, and done a bunch of work to make it easy for people to try.

ayyyyyyy

@argumatronic.bsky.social hiieeeeeeeeeeeeee

Imagine the world if this paper, published the year I was born (so, yeah, ancient), was more popular among programmers. Behold Programming as theory building by Peter Naur

Bicameral, Not Homoiconic Parenthetically Speaking: Articles by Shriram Krishnamurthi

@welltypedwit.ch what do you think of this bicameral syntax idea? I've found it incredibly helpful for learning Lisp, and for making sense of the rather confusing discourse that prevails in the Lisp world.
parentheticallyspeaking.org/articles/bic...

nice point; once you have static types, you can start to make your data structures suggest operations on them

riveting thread!

Stupendous book, please everybody buy it!

If you've thought about giving Haskell a try, this is a great opportunity to get Effective Haskell at a discount. Whether you check out Effective Haskell or not, I'm always happy to answer questions or help folks who are interested in functional programming.

*monadic

what, you mean you don't want a handrolled monaric parser from lists of characters into lists of parsetrees for your enterprise application?

wonder whether the utop devs would be up for a PR...

For the moment I'm settling for going back to utop mode with a plan to manually implement the comint features that I miss from doing it the old way.

right???? why the hell would a new compiler version bring the standard library out of scope??

Ang yet, although this definitely seems true in principle, I feel like I haven't figured out how to capitalize on it yet. Probably a skill issue.

Haha, yeah, I was going to mention, the fact that you can correct an LLM and follow up with it seems like the main contrast with Googling, where you're basically stuck with its first attempt to run your query.

For example, I asked both o1 and 4o about this and got absolutely nothing that was of any use.
discuss.ocaml.org/t/weird-glit...

LLMs are never able to help me with my computer problems, but I think that might be because my computer problems are too weird.