Pablo

Been programming just about everything for a long time.

[Hoyt Emerson @HoytEmerson]

This article from the DuckDB Team intrigued me. A DuckDB database does not need to hold any data, but instead view definitions pointing to Parquet data.

So it acts as a catalog for views that are run on object storage data and returned locally. No need to ever store data in the db itself.

https://t.co/enDu3GdzgL

[Tim Smart @tim_smart]

Native @EffectTS_ GraphQL client coming soon!

[Phil Nash @philnash]

Released a new version of https://t.co/RTbIyLAZ23 for verifying email addresses with the browser Email Verification API.

This version includes configurable timeouts for the network requests involved (after Google's JWKS endpoint was down last week).

[Ryan Dahl @rough__sea]

I, along with the rest of the Deno company, am joining Cloudflare!

I'm motivated by abstractions that give leverage at scale. The Cloudflare workers programming model is a hermetic computing abstraction that truly embodies the old Sun Microsystems tagline "The Network Is the Computer". The programming abstractions that Kenton designed really force scalability. If you squint your eyes you might even conceive of it as a distributed high-level operating system.

The only major problem with this distributed operating system is that practically people only use it inside of Cloudflare. Cloudflare is an amazing service for nearly all applications - but there are many situations when you might need it to run elsewhere.

I want to solve this problem and make the Workers programming model the default way you write server-side application code.

The time is ripe for new abstractions too - what software we run and how we run it are changing quickly with AI. The number of agents is exploding and I don't think they all need a Linux VM. A much more lightweight Durable Object would suffice for many applications. I'm super keen to talk to people who want to self-host agents at scale on DOs - msg me!

[Michael Arnaldi @MichaelArnaldi]

Your codebase is a form of memory, well put!

[egor @egorkabantsov]

Honestly, I didn't believe the hype first but I genuinely cannot go back to Claude Code or Codex after using T3Code. If you're not using it yet, what's wrong with you?

[Pablo @hadronomy]

@theo Loving it, but snoozing till reset is working really weirdly. It's coming out of the snooze like 30min later and the reset is like 2hrs and 30min later or something like that. Has happened a few times throughout the day

[dax @thdxr]

for the first time since the beginning of the year claude is our team's most used model

great work on opus 5.5

[Theo - t3.gg @theo]

"why do people take Theo seriously? he flip flops constantly between OpenAI and Anthropic"

Never understood this type of question. I use the best technology available to me at any given time. Switching cost is near 0. Why should I keep using worse stuff?

[Josh tried coding @joshtriedcoding]

one thing i LOVE about Opus 5.5 is that it consistently gets corner radii right

it just gets it right by default now, very impressive

[Matt 'TK' Taylor @MattieTK]

Today, Cloudflare have announced the intention to acquire a root cert and become our own CA 🎉 https://t.co/bj4ggRrJJh

[Matt Pocock @mattpocockuk]

Kill it - let users plan how they want

[ayush🔮👨‍💻🔮 @ayushagarwal027]

fearless_simd v1.0 is on its way: SIMD in Rust without writing unsafe. 🦀

By Shnatsel (Sergey Davidoff) and the Linebender team.

Most Rust SIMD abstractions rely on substantial internal unsafe code. fearless_simd concentrates that boundary in a small, audited implementation layer, so user-facing SIMD code can be written without unsafe blocks.

How:

The kernel! macro uses Rust's target-feature machinery and type-level SIMD capability tokens to make many architecture-specific intrinsic calls safe at the call site.

SIMD loads and stores ultimately operate on raw pointers. fearless_simd centralizes that handling in reusable safe-transmute wrappers, keeping the raw-pointer boundary in one auditable place.

Two small, auditable building blocks. The rest of your SIMD code stays safe.

What's also planned for v1.0:

Hardware-width vector abstractions, allowing one algorithm to target suitable 128-, 256-, or 512-bit implementations through dispatch.
→ Fast and precise variants for operations such as swizzle and floating-point maximum, whose behavior can differ across platforms.
→ Safe access to platform intrinsics when portable abstractions aren't enough.

fearless_simd_macros v0.1 with #[simd], providing function multiversioning without requiring #[inline(always)].

A planned three-year security-support commitment for v1.0 and later versions.

Even if std::simd stabilizes, fearless_simd can continue to provide value through multiversioning, hardware-width vectors, dispatch, and ecosystem tooling.

🔗 https://t.co/kr3EDtwz6o

#RustLang #SIMD #Performance #OpenSource #SystemsProgrammin

[Theo - t3.gg @theo]

Opus 5.5, GPT-6 Sol and GPT-6 Luna all in a day?? Damn

[Dillon Mulroy @dillon_mulroy]

i think jev is resonating with devs so well b/c it unlocks so many opportunities for composing ai into systems and products rather than ai _becoming_ the product/system

really does feel like it was a missing primitive

[Pablo @hadronomy]

I'm going to turn this into a poster and hang it on my wall

[Matt Pocock @mattpocockuk]

An actual slide from my talk at @aiDotEngineer Paris next Thursday.

People have this mad idea that their implementer agent should ALSO apply their coding standards

The solution? Hide your coding standards from your implementer agent, fix them in review

[Matt Pocock @mattpocockuk]

An actual slide from my talk at @aiDotEngineer Paris next Thursday.

People have this mad idea that their implementer agent should ALSO apply their coding standards

The solution? Hide your coding standards from your implementer agent, fix them in review

[Pablo @hadronomy]

@maria_rcks not having orchestrator v2

[Pablo @hadronomy]

@maria_rcks Switch codex sub mid thread, currently I have to restart the app before switching to avoid the thread panicking (I'm on nightly)