daenney

> We want to help people in the EU, but with laws like replaceable batteries, it's going to push us further and further away from being able to do that.

We want to help people, but only if and when it’s profitable for us to do so on terms we decide for you.

Incredibly corrupt sports organisation awards incredibly corrupt politician the peace prize. Something neither party have any concept of, nor seem to understand the definition of.

I feel for the South Park creators because boy do you have a challenge coming up with absurdity this season.

No. They do not.

There’s a lot of history there:

Rethinkdb is shutting down https://rethinkdb.com/blog/rethinkdb-shutdown/ (2016)

Why we failed https://gist.github.com/ramalho/93b87e961b6e019be8e1f6f82864...

It’s an open source project now and all but once all of that happened I guess a lot of people might not be comfortable to try it again.

“Millionaire steals from child.”

The title is interesting but the article lacks a lot of background information. There’s no explanation of what the CPU-bound endpoints are, what causes them to be CPU-bound etc. They mention they didn’t think optimising the Go code would’ve given them enough, but there’s nothing to substantiate it and no way for the reader to form their own opinion since we’re never told what the problem is other than “doesn’t scale.”

I have no issue with Rust, there’s nothing wrong with what they did, they approached it sensibly and the results are certainly compelling. But it reads a lot like “we wanted to write something in Rust” and found a reason to do so.

In the mean time, the author:

* Has claimed to quit Replit only to be back at it the next day.

* Asked the agent 6-7 times to generate apology letters.

* Is apparently spending $8000/month on "AI agents"

Like I'm not sure but it doesn't seem like this person should have any access to production either, much less the AI.

> While there’s no research to prove its effectiveness [..]

> Originally published in 1996, this article was republished on February 20th, 2025.

Ahhh the free market. In which we make sure to protect OpenAI’s profits.

It’s not landing gear failure.

They overshot the runway but that can happen for a variety of reasons. It’s nothing close to the other two.

There’s one useful paragraph of information in there, and the rest is Matt whining about the consequences of his own actions.

You can’t fix it. LLMs making things up is a consequence of what they are.

The blog post admits as much, because the “fix it” section is actually titled “How To Mitigate AI Hallucination?”

Mitigation makes some bad less severe. But that’s not fixing it.

File name doesn’t necessarily include the whole path. The last 16 characters of CHANGELOG.md is the full file name.

If we interpret it that way, that also explains why the filepathwalk solution solves the problem.

But if it’s really based on the last 16 characters of just the file name, not the whole path, then it feels like this problem should be a lot more common. At least in monorepos.

I’ve been very happy and impressed with the Framework AMD edition. I’d steer clear of their Core Ultra Intel edition since that’s Meteor Lake. I use mine for open source development things I do on my own time.

If work supplied me with one of them I’d happily use it. Support has been great and they live up to their upgradable promise.

This has been invented multiple times in JSON.

For a widely deployed example, take a look at how JSON-LD uses IRIs for keys and how the context object plays into that. The whole fediverse is built on top of it.

This is largely the same stuff Wired pushed.

The folks over at GrapheneOS have a much better analysis of this whole thing: https://grapheneos.social/@GrapheneOS/112967309987371034

This story is bogus. The vulnerability lives in a demo app that’s disabled by default. Enabling that app would require the attacker to have physical access to the device and your passcode.

The folks at GraphebeOS have a thread about it: https://grapheneos.social/@GrapheneOS/112967309987371034

The WASM solution doesn’t rely on a custom libc or transpiler to convert C code to Go. The transpile is an amazing feat of engineering, but it’s hard to debug.

I can wrap my head around the small amount of wrapping the go-sqlite3 WASM library does. If I had to I can maintain that should the maintainer lose interest. I can’t say the same for the modernc transpile. You can also apply the WASM trick to other libraries with much less effort.

And as noted, it seems to be performing better. As the wasm runtime improves it should pull further ahead.

Has been done and works pretty great: https://github.com/ncruces/go-sqlite3.

Though doesn’t convert the WASM to C, it runs the WASM in Wazero instead.

In the specific case of SQLite, you can use it through WASM now [1]. It uses the dependency and Cgo-free Wazero runtime.

Performance so far has been better than the modernc transpile and it’s probably sufficient for a lot of use cases.

[1] https://github.com/ncruces/go-sqlite3

The AI features wouldn’t have worked for the vast majority of EU citizens anyway. It’s “US English” only everywhere and while techies might use that a lot of people use their device in their native language, take notes in their native language, write emails in their native language etc.

The fact Apple believes their systemwide integrated features run afoul of the DMA is just a bonus.

> Today, the OpenAI Board formed a Safety and Security Committee led by directors Bret Taylor (Chair), Adam D’Angelo, Nicole Seligman, and Sam Altman (CEO).

A couple of directors, and the CEO. And a bunch of other "Head of" folks too. Might as well not have the Safety and Security Committee in that case.

This is unlikely to be of any consequence. ECHR has already ruled that weakening or backdooring e2ee isn’t allowed. So bar a law that bans rolling out e2ee at all, this won’t achieve anything.

If you want to elicit trust, you may want to consider eliminating all the trackers and other stuff you have on the site

> We use cookies to personalize content and to analyze website traffic. We also share information about your use of our site with our social media and analytics partners, as noted on our Privacy Policy. If you continue using this site, you consent to our use of cookies.

> By accepting all services, you allow NitroPack, Twitter (embedded tweet), YouTube, Google Fonts, Google Tag Manager, HotJar, google analytics, Linkedin, Hubspot Analytics, Popup Maker and Facebook Pixel to be loaded. In addition, cloudflare CDN, JS Deliver CDN, WPML, Cloudflare and Wordfence are loaded based on a legitimate interest.

The submitted link results in a login wall. It would probably be worth updating the submission to the introducing-dhai page that's mentioned all the way at the end here, as getting greeted by a login prompt that way isn't going to entice people to explore or learn about the product.

As usual, these articles are complete garbage. I can only assume at this point that they wilfully conflate the highest tax brackets with the effective tax rate since that makes for a more dramatic story.

Yes Sweden has a tax bracket that would result in north of 50% tax. For the part of your income that qualifies for that bracket. A significant part of the population doesn’t even qualify for that bracket and for those that do it’s a much smaller part of their total income.

If you earn so much that the dominating tax bracket for you is the highest one, good. System works as intended.

No, it’s an English term. See for example https://www.merriam-webster.com/dictionary/fair

Is that what we're calling missiles now?

Yes. In their JS, they check if you're coming from HN:

      if (document.referrer) {
        const ref = new URL(document.referrer);
        if (ref.host === 'news.ycombinator.com') {
          window.location.href = 'https://google.com/';
        }
      }

You can copy-paste the link instead and it loads as expected.

No, that is not a correct assumption. It's also worth remembering that cloud providers are also servers in racks, but they own the building around it to.

For DDoS attacks, you need to have enough capacity to absorb the attack. Major cloud providers tend to have that, as do DDoS mitigation services (Cloudflare amongst others).

> We spoke to CloudFlare and were quoted a number we cannot reasonably achieve within our financial means [..]

Typically what you want to do though is stop the traffic from reaching you at all, so ideally your network provider, who is upstream from you, blocks the illegitimate traffic so your servers never see it and don't get overwhelmed.

What happened here is that due to some administrative lapses, the victims (Sourcehut) of the attack got disabled by the network provider. That was the initial outage. Imagine if your ISP decided to stop routing traffic to Google. Being hosted on GCP, a major cloud provider, would be of no use, since there wouldn't be a network path to them in the first place.

In general, Cogent seems to be doing a rather bad job at dealing with this attack and there's been fallout for many services beyond Sourchut. Google or AWS or Microsoft might've handled it more gracefully, or might not. Though major cloud providers tend to have their own connectivity between their datacenters, they too have peering/transit agreements with other major network providers. If those upstreams stop forwarding traffic to them, the same thing would happen. It's just less likely to go unnoticed.

Cogent is a massive provider, so you'd think they'd be a bit better at this. But they also have a reputation for being awful.