rattray

I think that gets you access to mythos, which doesn't have the safeguards. It's configured as a separate model.

GP seemed to be commenting on the Trump administration, not necessarily individuals of conservative persuasion. The Trump administration diverges materially from traditional conservative doctrine in many ways.

... and their team is joining Anthropic to build, who knows, some dumb integration to make Hubspot data available in Claude, or something equally as boring.

Fun fact, I named it "Stainless" after Stainless Steel pipes, likening ourselves to a high-end plumbing supply shop. If you look at the earliest versions of stainlessapi.com on archive.org, you'll see our original motto was "Quality fittings for your REST API".

All that is to say, the incredibly "boring" infrastructural work of making "boring" APIs like Hubspot's more usefully accessible is absolutely the kind of thing I'm excited to do at Anthropic :)

(It also happens to be what got us all excited to work at stainless in the first place, but of course, we understand it's not for everyone!)

Aww, well this thread is a nice surprise :) thanks for the kind words!

TypeSpec is awesome!!

(disclaimer: founder of Stainless and also friends with creator of TypeSpec)

darn! anything else i missed?

Hmm, that's one explanation, but I'm curious what leads you to believe it's the correct one?

I'm struck by this quote, which I'd be surprised if they could be explained fully by the distraction-reduction mentioned:

"Evaluation of symptoms on the Parent Global Impressions found that the treatment group at the end of part 2 improved more than the placebo group in part 1 on nearly all symptoms, with statistically significant improvements in GI, receptive language, and average of all symptoms. There were also marginally significant improvements in tantrums, stimming/perseveration, and cognition."

Prior to the study, 83% of participants had "severe" autism. Two years later, only 17% were rated as severe, 39% as mild or moderate, and incredibly, 44% were below the cut-off for mild ASD.

Pretty incredible if true!

why not just build a translation tool for it?

They did ;) a highly dynamic one...

It's possible that use of `contenteditable` and ability to save the file could help but that has a lot of limitations/gotchas, so I'm inclined to agree.

Cool! How'd you build it / what made it work so well?

Aren't there high quality cables available? Eg from companies like Apple or Anker?

I struggle to understand why an Apple lightning cable would be more robust than an Apple thunderbolt cable...

https://archive.ph/U7i8q

I've been very curious about that too. I wonder if it's actually much better at admitting when it doesn't know something, because it thinks it's a "dumber model". But I haven't played with this at all myself.

Has anyone tried both jj and gitbutler extensively yet? Both seem to have some interesting new ideas on top of git.

What's your naming scheme?

Abandoned open-source projects with poor code quality are nothing new.

The merits of any project are yours to evaluate.

To me, I see some encouraging thoughtfulness here. However, again, it's true most projects like this don't achieve liftoff.

This seems awesome. Seems to address many of my armchair complaints about both Go (inexpensive) and Rust (bloated/complex).

I'm curious what compilation times are like? Are there theoretical reasons it'd be order of magnitude slower than Go? I assume it does much less than the rust compiler...

Relatedly, I'd be curious to see some of the things from Rust this doesn't include, ideally in the docs. Eg I assume borrow checking, various data types, maybe async etc are intentionally omitted?

Some details on how it works from a code comment:

Problem: DOM-based text measurement (getBoundingClientRect, offsetHeight) forces synchronous layout reflow. When components independently measure text, each measurement triggers a reflow of the entire document. This creates read/write interleaving that can cost 30ms+ per frame for 500 text blocks.

Solution: two-phase measurement centered around canvas measureText.

prepare(text, font) — segments text via Intl.Segmenter, measures each word via canvas, caches widths, and does one cached DOM calibration read per font when emoji correction is needed. Call once when text first appears.

layout(prepared, maxWidth, lineHeight) — walks cached word widths with pure arithmetic to count lines and compute height. Call on every resize. ~0.0002ms per text.

https://github.com/chenglou/pretext/blob/main/src/layout.ts

Regardless of the subject matter, the tweets announcing this are a masterclass in demoing why an architectural/platform improvement can be impactful.

Wow. How fast is haiku?

For our first experiment, we used ClickBench, an analytical database benchmark. ClickBench has 43 queries that focus on aggregation and filtering operations. The operations run on a single wide table with 100M rows, which uses about 14 GB when serialized to Parquet and 75 GB when stored in CSV format.

very much so…

Curious, what do you use it for?

Does anyone have a link to good video footage of the climb?

Reading the code felt a little eye opening about what programming can be. Slightly like learning a bit of Haskell. Even though sql is so familiar.

Does capn proto have similar properties?

Ah, in that context, why not just give the people workerd? People using & running OSS libraries are used to the fact that there might be vulns in libraries they're using, right?

Wait, why not just actually use the Cloudflare Sandboxes product then? Is it too costly or something? Or you need to be able to run without a connection to their cloud?

https://developers.cloudflare.com/sandbox/

We've been looking at using this at Stainless. Our evaluation isn't complete but my personal gut feeling is we'll be incredulous that we ever operated without it.

is there a way others can contribute their favorite motions? like a link to the relevant bit of source that could be shared here?