Pinned Tweet
We just built the most scalable Git server out there: 7,600+ concurrent pushes/s per repo at millisecond latency.
Powered by Raft and S3, designed like a modern database.
Today we're making it open source by releasing Enroute 0.3.0.
Read the write-up: enroute.sh/blog/everyone-is-…
Sam Reis retweeted
isn’t it beautiful the competition that ai has enabled?
we have people trying to build more scalable git, postgres, sandboxes… the result of that competition will benefit our entire industry, and this wouldn’t have been possible years ago
Whenever you‘re dealing with Git objects, (de)compression shows up big time in cpu profiles.
I just open sourced sirocco (github.com/srijs/sirocco) which for avg git pack parsing is now the fastest streaming inflate lib that I know of.
Ships in our next release.
This is 💯% the thesis we have at @enroute_sh.
The companies of the future will be made out of code; small apps, scripts, the remotion project that generates your next slide deck.
And we need the infra that supports it at that scale.
Notion AI lead @sarahmsachs argues that a lot of what runs on GPUs today should actually run on CPUs:
"We have a strong perspective at Notion that a lot of things should be served on CPU that are served on GPU. There’s a lot of benefit and determinism in executed code."
"Agents don’t need LLMs to reinvent that process every time, and that’s super token exhaustive, it’s super slow, and there’s no observability or determinism in it."
"Giving the agent a local file system to store local thoughts, to write short Python scripts, that’s really important when there’s so much information. That has been probably one of the most important tools outside of edit and search that our internal agents rely on."
@NotionHQ
Great to see more people thinking about bring distsys techniques to scaling git on the server.
We have been working on our own version of this, will release soon. Stay tuned for the blog post!
Sam Reis retweeted
We are verification engineers now. With the cost of code going down, we have to reassess the trade-offs of technologies that make coding relatively more expensive but increase robustness.
What might have been too much of a hassle in practice in the past, is now worth it.
We've since undertaken the mission to make IDOR bugs much less likely on our platform using proofs enforced by the typesystem.
Introducing gdp-ts: Ghosts of Departed Proofs for TypeScript.
gdp-ts is a library, linter and AI skill for safer API design. Under this contract, sensitive functions require 'proofs' that the caller performed an authorization check.
The typechecker verifies these proofs at compile time, preventing your team and agents from shipping catastrophic security (and other kinds of) bugs.
While these patterns have existed for quite some time, especially in ecosystems like Haskell, ① human code review and ② cognitive and syntactic overhead made these solutions niche.
The situation is now inverted. Agents are writing more code than we can review, and they *thrive* in tight loops with hard constraints that would frustrate us. We see this with the rise of Rust, borrow checker, code aesthetics debate and all.
The README and examples model a real-world Vercel API product constraint: changing the password on a Project requires a proof of a certain role + a certain entitlement. Thanks to Matt Noonan and Ollie Charles for their research in this space.
github.com/rauchg/gdp-ts
Happy to be included in the September release of Gitoxide 🚀
github.com/GitoxideLabs/gito…
Fast git packfile verification coming to #gitoxide.
github.com/GitoxideLabs/gito…
I just wrote a new collision-detecting SHA-1 hasher in Rust that runs almost 3x as fast as existing options thanks to SIMD and which can 2x+ the performance of git pack verification.
It's on crates.io as sha1dc.
Check out sam.dev/blog/faster-sha1-col… for how it works!
Sam Reis retweeted
If you work with Tokio, this is a must read: "Principles for fast Tokio applications" dial9-rs.github.io/blog/prin…
Effect systems in the frontend coming back about a decade after PureScript definitely wasn’t on my bingo card, but it sure is interesting to see how coding with LLMs is shifting trade-offs. Feel like the same is happening with Rust and Nix, curious what’s next.
My gearhash crate is now pulling 10-20k downloads a day, thanks to powering @huggingface's Xet client, so it felt about time to revisit the code.
Just published v0.1.4: ~2× faster on ARM64 🚀.
Check out this post for the fun details of how it got there:
sam.dev/blog/gearhash-on-arm…
Me: Asking for a custom favicon.
Claude, 30 seconds later: "I wrote a small pure-Python PNG decoder/encoder in the scratchpad to do it."
Just crazy how much we're taking for granted these days.
This 💯. Almost all my direct effort on code reviews and building guardrails is currently going into keeping complexity contained.
All models, no matter how smart, will eventually build systems that they can no longer understand or maintain, if you let them. Fable 5 finally outbuilt itself, and flailed on me for a week. Fable 5.1 looks like it will fix it. For now. But you have to keep an iron grip on system size, or it'll run away from you.
Sam Reis retweeted
✨ Wrote about building and growing Internet services with feature boundaries, data autonomy, cell-based architectures, service dependencies, graceful degradation, unique identifiers, idempotency, SLOs and quotas, and observability. rakyll.medium.com/how-to-pea…
Sam Reis retweeted
We’re brewing something hot 🧙♂️🔥 ..
> “Allow your users, team & 3rd party developers to extend your product’s functionality + integrations in minutes. 🚀”
Let me know if this sounds interesting and I’ll get you on our “secret” early access list 😎 First come first serve. 🍦🍫🍬
Sam Reis retweeted
I had so much fun giving my talk about apollo-rs and its compiler! Thank you so much for having me #rustconf.
(It's been a while since I've done this, but!!!) here are my slides ✨
slides.com/lrlna/writing_a_g…
