gajus

As a long-time open-source contributor, I've always been fascinated by using GitHub to track trends. However, recent GitHub changes (whatever the public stated reasons are) make parts of what used to be public data unavailable and raise concerns about the future of other data availability. Similar to Internet Archive, I think it is important to preserve public GitHub data.

If you are researcher with the need to access the data or would like to donate to the project, get in touch.

If you are curious how it compares against StyleX and Panda, I made a public benchmark https://github.com/gajus/css-in-js-arena

There is a lot of jargon in the title, but the key things to know about Bamboo are that:

* it eliminates all runtime overhead for static styles

* it pre-computes all dynamic styles leaving only boolean logic in runtime

* parsing all your source code is computationally expensive, therefore parts of Bamboo are written in Rust and leverage Oxc to make it fast

Happy to take any questions

The funny thing is that DO itself has an MCP for spinning up VMs, resolving IPs, etc.

OpenRouter is valuable because it has a large catalogue of competing providers for a fungible service. This does the opposite (lock in with a specific vendor).

Can you give me a practical example of 3 most deserving use cases? Something that customers are actually using them for today.

A few other things to callout:

* Bamboo started as a fork of Panda. The vast majority of the codebase has been refactored. However, their test suite has been incredibly useful to identify all the edge cases.

* Bamboo's API is a subset of Panda. One of Bamboo goals has been to introduce more consistency to how we write styles (without introducing lint rules). Bamboo dropped JSX factories, template literals and style props.

* Strict API is also what allows us fold 100% of styles.

* Strict API is also what allows to prune unused styles.

* Bamboo resolution is _strict_. Missing tokens will produce build error.

this gives strong "you wouldn't steal a car" vibes

The moment Google announced SynthID, the first domain I bought was deSynthID.com

Several open-source projects have already proven SynthID to be ineffective.

There are already small models trained specifically to prevent statistical detection, e.g., https://huggingface.co/kalpeshk2011/dipper-paraphraser-xxl

I guess whoever is the policy maker is assuming that some protection is better than none and that most people will not reach for such tools.

Most likely watermark will be proportional to the input/output ratio, i.e. if you input a long document and ask to make edits, it will not attempt to watermark it. On the other hand, if you provide a tweet and ask it to write an article, that will include watermark. Just a guess (and yes, it feels flawed)

so... what happened?

Excellent problem space.

Related problem: Using AI to write compilers that optimize runtime code. I did this for Zod.

https://github.com/gajus/zod-compiler

Zod schemas, compiled at build time into raw boolean chains. 2-74x faster. No code changes. The plugin swaps Zod with compiled parse.

I wrote almost none of the optimizations. Claude did, in a loop, over 100+ iterations.

It works because the compiled version can be diffed against real Zod. Correctness isn't a judgment call. Same as scriptc.

This applies to every program with a reference implementation and a benchmark. Compilers, serializers, formatters, query planners, etc.

Excited for faster Internet.

What a throwback. Back in 2015 I have started Applaudience, which at the time was the only provider of real-time cinema ticket sales data. I still remember comparing our numbers against TheNumbers.com as part of calibration. I have since moved on to other businesses, but this remains one of my favorite pieces of technology that I have developed. Would love to bring it back one day.

Reminder to secure your npm environments.

https://gajus.com/blog/3-pnpm-settings-to-protect-yourself-f...

Just a handful of settings to save a whole lot of trouble.

A reminder to secure your npm environment:

https://gajus.com/blog/3-pnpm-settings-to-protect-yourself-f...

Congrats to the team!

We've been using it (in beta) for months now. It makes for a very enjoyable developer experience.

Funny you ask. This project started as a very different project almost five years ago. It was called roarr.io, and the primary purpose was exactly that: adhoc collecting logs from remote machines. However, I've not ported this functionality (yet).

That would require restarting your services to redirect their output. Fine for one-off scripts, but impractical when you have long-running processes and don't want to restart them every time an agent needs to read logs.

With teemux, a persistent MCP server gives multiple AI agents access to logs as needed—without interrupting your development flow.

Fair point

lowkey thought it is a genius name

tee (Unix command that splits output) + mux (multiplexer) = teemux

Will be interesting to see how this model performs in real-world creative tasks. https://creativearena.ai/

Man, we are living in the golden era of PostgreSQL.

The migration guide https://vitest.dev/guide/migration.html#vitest-4

Somewhat related to your ask https://github.com/gajus/pg-dump-parser

We use it internally to create folder-like structure representation of our database schema.

https://github.com/gajus/pg-dump-parser?tab=readme-ov-file#r...

Very handy when reviewing changes in version control, etc.

Zod is the default validator for https://github.com/gajus/slonik.

Zod alone accounts for a significant portion of the CPU time.

Surprised not to see more people ask about performance profile of v4.

Zod is great in terms of API, but a no-go in terms of performance.

We ended up writing babel plugins (https://github.com/gajus/babel-plugin-zod/) and using it along with zod-accelerator, which improves performance, but breaks in various edge-cases.

What's a relatively small dataset?

For someone that could be 1m records for others that could be 1bn records

TLDR

> So, we now focus on the new NIST standards of FIPS 203 (ML-KEM), FIPS 204 (ML-DSA) and FIPS 205 (FLH-DSA).

That’s a lot longer post than I anticipated on the subject.

Just so I understand the other side of the argument: where is the purported inefficiency of keeping DST coming from?