Joined June 2011
- Tweets 3,879
- Following 205
- Followers 303
- Likes 3,022
204 Photos and videos
Jun 9
Reading an agentic PR I am the owner of at this point is like reading another person's PR, the only common thing I have with the deep understanding of the PR like I used to with my own code is it has my name assigned to it, and a task that I know of.
1
21
Itay Adler retweeted
Jun 3
Elixir v1.20 released! Now officially a gradually typed language: Elixir type checks every single line of code, finding bugs and dead code, without developer overhead (no typing signatures) and extremely low false positives rate. Plus a faster compiler! Links and reports below.
56
307
1,466
116,393
Itay Adler retweeted
May 31
i'm really curious to let AI refactor our monorepo build scripts to Bazel.
i mean. the only reason not to use Bazel(or similar) previous was the manual work that was just extremely hard to do and maintain.
but with LLMs?
1
2
133
Itay Adler retweeted
May 29
Used the LLM as a guide today, not a ghostwriter.
It felt exactly the same speed as agentic coding.
Just now i can reason about the system.
2
1
4
712
May 27
David Firth predicted the AI revolution before everyone youtube.com/watch?v=0UgiJPnw…
1
2
188
May 26
how many times do you tell Claude/GPT to review and fix critical issues it finds before reading the code? also lately I like threatening to fire it if it doesn’t produce the same functionality in much less LOC
55
in Riot, one of the things i wanted to change from how OCaml does package management was the registry: pkgs.ml
so pkgs.ml was born as a little registry service. you login with github, you create a publishing token, it tracks package, riot, and ocaml install metrics, etc. you can go to pkgs.ml/stats and see some of the metrics being collected there
you can also see all the publishing activity in pkgs.ml/activity
and ofc, since it collects data,there's pkgs.ml/privacy! don't worry, we only collect a handful of things that are useful insights for package authors ;)
These last 6 months i've been building a tech demo showcasing what I believe OCaml, its ecosystem and its DX could grow into.
I started with the Riot ecosystem I had been building a few years back (scheduler, tui, web framework, etc), and then I added one tool, and another, and then it got out of hand.
The end result was a complete platform that I genuinely enjoyed using daily to build all sorts of tools, almost all of it AI-assisted! It's a pretty great way of building with OCaml, even if its just a tech demo :)
riot.ml and pkgs.ml
No camels or dunes were hurt during the development of this tech demo, this isn't official OCaml stuff, use at your own risk, etc.
Sources on github.com/leostera/riot <- go to drop a star! 🌟
1
1
9
476
May 25
hmm why are the most popular posts in my feed are always “which is better, this or that?” every day it’s the same god damn thing.
youtu.be/9EYZnSXEla0
1
65
May 25
really appreciate @adamdotdev sharing his AI psychosis arc, I feel I had a similar experience at around jan-feb this year, with a fallout that got me to stay away from my main project for 1-2 months
we need a human harness to control the agent harnesses
youtube.com/watch?v=Enbqwpkm…
1
5
3,199
May 25
is there any agent harness that spawns subagents for different models and allows to route based on some config?
2
37
i've been thinking a lot about how to encode repeatable quality checks for my agents without spending a lot of energy/effort/tokens into making custom lints for different linters, and without it being too slow to execute!
cons for just using agents here is:
* non deterministic, might miss checks!
* slow, boy i need to save me some ms and one llm call per group of files is just not scalable
* expensive, tokens burn like crazy if you use it on every commit
and so lintbook was born: author lints with llms in simple .md files, execute generated datalog queries
* deterministic: its just datalog queries!
* fast: a ~30k project runs in ~300ms
* cheap: only uses tokens when compiling lints
github.com/leostera/lintbook
1
4
21
1,619
May 22
at the end of this we will find out this was a commercial for a new high end service for people, clone as a service
1
63
Itay Adler retweeted
May 20
Fork your dependencies, trim them to only your use case, never update unless it breaks for your users. I’ve been vocal about this for 10 years. I’ve always said that updating is way riskier than latent bugs (which can be tracked and CVEs monitored).
If you are updating a dependency, it’s on you to analyze every single commit in the full transitive set of dependencies. If you dont see anything compelling, dont update!
I remember at HashiCorp once in awhile an engineer would try to update a dep or replace a DIY lib with an external one and id always ask “show me the commit we need.” Dont update for the sake of it.
Feeling pretty swell about this mentality with all the supply chain attacks happening.
292
776
8,956
1,184,703
Itay Adler retweeted
May 18
commercial break, we also have some technical posts
Feb 19
1/ AI code editors parse your files but never attach to the runtime. that's like working on a car engine by reading the owner's manual.
here's what you actually lose:
2
1
194
May 18
we know bro
May 18
ChatGPT Images 2.0 💚 India.
Already more than 1 billion images created there; awesome to see.
1
105
