Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
the8472
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
39 ms
·
691.
▲
by
the8472
3y ago
Much simpler: They'll make WEI-verification mandatory for all government sites. And then banking. And then anyone doing commerce.
692.
▲
by
the8472
3y ago
On the other hand, old software often runs better (or at all) on dosbox or wine than on windows.
693.
▲
by
the8472
3y ago
Well, ok, let's assume for a moment that ads are valuable to some users, perhaps informing them of a solution to a problem they didn't even know had solutions. That still wastes the attention of everyone who doesn't actually
694.
▲
by
the8472
3y ago
It would be possible to make iterators more SIMD-friendly than they are today but that would come at the cost of interpreting what the user might have intended and then dropping or introducing additional side-effects (e.g. by skipping thing
695.
▲
by
the8472
3y ago
> free software updates It's not like amazon is paying for the maintenance of the entire software stack. And the part they're doing they mostly have to do anyway to support new devices.
696.
▲
by
the8472
3y ago
Loading 100GB into RAM and then calling fork() is just painting a giant OOM Killer target on your back. It'll work until something breaks the CoWs or the parent gets restarted while some forks still linger or other fun things like that
697.
▲
by
the8472
3y ago
> I thought Single threaded execution without overhead for concurrency primitives is the best way to high performance computing You can have shared-memory parallelism with near-zero synchronization overhead. Rust's rayon is an examp
698.
▲
by
the8472
3y ago
Threads are cheap compared to the process-per-request model that came before. And they're easy to code for, you don't have to worry about blocking code or awaiting your futures in the correct way or things like that. They'r
699.
▲
by
the8472
3y ago
The ancient server box doesn't need to do that. Publishing the hashes to DNS can happen somewhere else. And the clients would have to trust their local resolver if they don't want to do the validation themselves.
700.
▲
by
the8472
3y ago
> How do you open a private channel with the website ? I wasn't assuming you would. That's one of the things that have to go on the chopping block if long-term stability is the goal.
701.
▲
by
the8472
3y ago
I haven't mentioned X25519 anywhere.
702.
▲
by
the8472
3y ago
The TLS treadmill + certificate authority lists are antithetical to stability. HTTP + merkle-tree based integrity checking + having the merkle root in DNS could be way more stable. Even the the SHA1 deprecation would barely affect this, as
703.
▲
by
the8472
3y ago
I would guess we'll see steeper gradients which means more extreme weather.
704.
▲
by
the8472
3y ago
clone/exec via io_uring will hopefully end this madness. https://www.youtube.com/watch?v=jrA-y1HG31E https://lpc.events/event/16/contributions/1213/attachments/1... https:
705.
▲
by
the8472
3y ago
> But says that the data thus far had not borne out global warming was happening. Should we wait until predictions of bad outcomes become true and only then react? Has anyone tried adding a D part to the PI policy.
706.
▲
by
the8472
3y ago
> We don't expect Linux to do this, do we? Linux can be setup to verify an immutable userland via secure boot + dm-verity. I think it's used in some google devices (chromebooks?).
707.
▲
by
the8472
3y ago
> But what's the alternative ? Since ads are a red queen's race, wasting people's attention and assuming we don't want to outright ban them then the next best alternative is to tax them heavily. How about 0.00001 cent
708.
▲
by
the8472
3y ago
But that restriction also makes it incompatible with other licences which means the code can't be used by other FOSS projects. And often people care more about that. It'd be great if we had something like an LAGPL though.
709.
▲
by
the8472
3y ago
I think most people don't want to be that anti-corporate. Otherwise they wouldn't use permissive licenses in the first place. And there may be non-license options that are friendlier to open source than corporate. E.g. if someon
710.
▲
by
the8472
3y ago
Depends, how do you distinguish humans acquiring knowledge by ingesting copyrighted content vs. a human using an AI that ingested copyrighted content?
711.
▲
by
the8472
3y ago
https://gwern.net/scaling-hypothesis#why-does-pretraining-wo...
712.
▲
by
the8472
3y ago
That's not really specific to the language barrier though. Anything discussed verbally will be lost to those who are not present, irrespective of discussion language. Ticket with motivation, proposed solutions, considered alternatives,
713.
▲
by
the8472
3y ago
Humans are also trained on copyrighted content they see. Should every artist have to pay that fee too on every work they create? Disney will finally be able to charge a "you know what the mouse looks like" tax.
714.
▲
by
the8472
3y ago
> perhaps there will be a flag to selectively operate a python process without the GIL (and leave the onous on you to completely test and validate your code works with potentially new and spooky behaviour). Worked for ruby. The original
715.
▲
by
the8472
3y ago
https://en.wikipedia.org/wiki/Information-theoretic_death
716.
▲
by
the8472
3y ago
That seems unlikely to me. Models are multi-modal now, so they'll likely train on video and audio content too. Once that's tapped out they can learn better world models by not passively ingesting data but interacting with it, e.g.
717.
▲
by
the8472
3y ago
Any copy command. On-demand deduplication managed by userspace. https://man7.org/linux/man-pages/man2/ioctl_fideduperange.2.... https://man7.org/linux/man-pages/man2/copy_file_r
718.
▲
by
the8472
3y ago
Could the linux integration use dax[0] to bypass the page cache and go straight to ARC? [0] https://www.kernel.org/doc/Documentation/filesystems/dax.txt
719.
▲
by
the8472
3y ago
You get a recent kernel + mesa + vulkan which means a recent proton shipped with steam can use all the extensions it wants for efficiency. Recent kernels also contain a few features specifically designed to aid wine. Also, it's probabl
720.
▲
by
the8472
3y ago
> All memory would be prepared in advance and the massively parallel accelerator would "stream walk" through It without having to load store arbitrary memory locations. Dedicated accelerators do that already, e.g. google's
More ›