mkj

openai etc are going to have a higher utilisation of the hardware so can afford it more than small companies/people. Efficient resource use matters more when they're expensive.

Interesting idea, but wouldn't an embedded language normally be interpreted/runtime editable? I guess you could do it shipping a rust compiler, but it'd be big. A bit like Numba?

getaddrinfo was added after ipv6. Software had to be rewritten to use getaddrinfo.

Prior to that programs used gethostbyname() etc, which only works with ipv4.

Old Nanostations as a client need to do proxy arp or something, which doesn't handle ipv6. That said it's probably 15 year old hardware. I ended up using a wireguard tunnel across it instead.

Maybe the LLMs need some kind of "coverage" metric so they prioritise new paths? The author would know a thing or two about that.

I think that's incorrect. Brains/eyes are good at looking for patterns.

If you have less than hundreds of thousands of points, you can probably get away with drawing individual dots on the map. Computer screens are high res.

There are excellent examples of the problem in Edward Tufte's "Envisioning Information". Have a look at http://blah.ksteinfe.com/181106/tufte_envisioning_informatio... for a few snippets of it, then buy the book.

I'm wondering who's going to buy pi!

(Edit sorry forgetting names, I mean who's going to buy Earendil). Good luck to Armin, he's done some good stuff.

It looks like github is just buggy in this case.

From what distance? I would have thought $1M wouldn't go far

Wait until they hear how much source code is thrown away to make software! I guess that's a bit ephemeral, but still probably similar in a lot of industries.

Clothes are weird shaped, weaves are rectilinear, it's a pretty tricky problem to solve. Unless someone manages to invent a non-rectilinear robot loom or something?

Solvinity is a pretty terrible company name.

"needs clarification" looks like a pretty normal hardware design process as the details are finalised.

https://github.com/flipperdevices/flipperone-docs/commits/pu...

The rustcrypto crates seem a lot better maintained and will probably supplant ring in future. Ring maintenance seems hostile to use outside of expected use cases.

If they're already running a custom Linux kernel build, why did they have AF_ALG enabled? Seems the perfect situation to limit features to only those actually being used.

They're talking about billions of dollars of market share, so how does debian get a mention being free? I'm suspicious of their methodology.

At least the infographics down the bottom are obviously full of slop

Tokio’s dominance is function coloring at ecosystem scale

That isn't function colouring, but rather plain incompatible APIs/runtime. You could have the equivalent with non-async ecosystems.

How is docker a context switch overhead? It's the same processes running on the same kernel.

What does ed448 mitigate against vs ed25519?

Fair enough. A 15 commit PR is still pretty long winded.

Looking at the PR discussed, it's 34 commits! I'd probably ignore that too as a maintainer. The PR description isn't particularly motivating, "Cleans up the implementation", "see #6735 for the actual motivation".

Isn't Nix just reinventing what Vesta did for software reproducibility decades earlier? https://vesta.sourceforge.net/

Are you saying Bram hasn't worked on VCS problems much? https://web.archive.org/web/20071213090008/http://codeville.... is 20 years.

It looks like firecracker already supports ACPI vmgenid, which will trigger Linux random to reseed? https://github.com/firecracker-microvm/firecracker/blob/main...

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin...

So that just (!) leaves userspace PRNGs.

BSD0 doesnt

Does that page even say which RISC-V CPUs are being used that are slow? I couldn't see it, which seems a bit of pointless complaining.

Intriguing work! Does it panic on any bad inputs? That's better than memory unsafety of libxml2, but still a DoS concern for some servers.

No it's not, it's 6pm!

For example, std::time::Instant is implemented on the GPU using a device timer

The code is running on the gpu there. It looks like remote calls are only for "IO", the compiled stdlib is generally running on gpu. (Going just from the post, haven't looked at any details)