What is the logic there? How do you get from "police have misused it" to "police should be banned from using it"? I feel a few steps are missing.

Easy - the justice system isn't doing the proper checks. Police use it, and it returns a list of names of people WHO LOOK NOTHING LIKE the person of interest. And police do not even perform a preliminary check of whether or not the person could have committed the crime (they may have an alibi). And the judges do not even take 5 seconds to look at the photos of the suspect or the person identified. And then you arrest a random stranger who is then locked up for several weeks, ruining their lives (lose their job, lose their house, etc).

No, they're not supposed to be using the software and going from "surveillance video" to "arrest"

It should not take a judge in a courtroom to have to dismiss the case for obvious "this guy should never have been arrested in the first place".

It's only a matter of time before facial recognition says a black guy did the crime when it was clearly a white guy in the surveillance video. Or vice-versa.

And as long as police do not use their tools properly they shouldn't have access to them. And using the tools properly means understanding the limitations of it. Facial recognition is not a magic box where you insert video on one end and it spits out the name of the culprit on the other.

No, facial recognition is a tool, and it can be used, when used responsibly. The problem is, the police have not shown to be responsible users of the technology.

Especially when they can ruin lives due to misidentification. Once arrested, it can take 2 or 3 weeks before anyone processes your case. Which means you're stuck in jail for 2-3 weeks. Your job will likely fire you for not showing up to work. And you can lose your house because without money, no job, there goes your rent or mortgage payment. 21 days later someone goes "Oops, you look nothing like the suspect, sorry!" and then they dump you on the street. You lost your job, your house, your family is likely somewhere where you don't know.

(And yes, ICE uses the same tactics for those people they wrongly arrest)

If you're lucky great, you can get a lawyer who can expedite matters - but that shouldn't be the determining factor on whether a false arrest ruins your life.

time_t has been 64 bit on Linux for a long time now (over a decade). You haven't needed a 64-bit system for 64-bit time_t.

x32 changes the model for 64-bit computing - these are the C semantics.

Remember in C, sizeof(int) = sizeof(long) = sizeof(long long)

You can have 32 bit ints, 64 bit longs (long longs are 64-bit), and 64 bit pointers - referred to as ILP64 (int, long, pointer). This is traditional in Linux and similar operating systems.

Or you can have 32-bit int, 32-bit long, and 64-bit long longs (and 64-bit pointers) - LLP64 (long long, pointer) - this is Windows where lots of code assumed ints and longs were 32 bit.

x32 is where int is 64-bits, longs are 64-bit, but pointers are 32-bit. The code runs in 64-bit mode (on amd64 systems, this has benefits because amd64 has access to many more registers than in ia32), but pointers and address space is 32-bits (4GiB max for the application - note the kernel does not have to live in the upper memory).

It's a weird mode of operation because it's based on something where having the extra wide register set but limited memory space is a benefit, or for some architectures like ARM where AArch64 is much more efficient. Likely this is for operations where the data to be worked on is wide but streamed - think DSP or other signal processing applications where you don't need 64-bit pointers because your code is tiny, your data is moved back and forth between several buffers, but would like 64-bit wide registers to process with and having a lot of them is useful. Having 64-bit pointers would just be a waste.

But it's really a niche application where you're having so many pointers it might have an impact on RAM usage if you needed to keep them 64-bit.

This round of consoles is an oddity - the PS5 and Xbox Series have upended the usual discounting rules, and they cost more now than they did in the past. Usually by this time the price of new consoles would've decreased by 50% or more (remember, we're 6 years into the console lifespan).

Now the Steam Deck costs more than what I paid for it.

Though, at least I can say, I got my money's worth of play out of the Steam Deck - it suits my needs for portable PC gaming very nicely.

Yeah, they should've started with something much more reasonable first - like an automated reader that catches people who drive past school buses that are stopped picking up or dropping off students.

Blowing past a school bus with its stop lights on and the sign out is illegal in most jurisdictions and can be implemented fairly easily.

You do that first, which gets the cameras mounted and recording continually standard on all school buses. Then you implement them to do all license plates later on once everyone is used to them.

That's how you do it - you get people onboard with it first then you expand its capabilities.

For space systems, there are a few choices.

For ground based gateways, there are a lot more choices.

For domestic flights, most in-flight WiFi is ground based (think a really sophisticated form of cellular technology). This is because you only need a few ground stations in various locations in the US for coverage. It's also much cheaper than space-based systems, which is why most domestic flights have made in-flight WiFi basically free.

It's also why international flights the in-flight WiFi is often horrendously expensive.

The real innovation here is that Starlink is relatively cheap enough that you can offer in-flight WiFi on domestic flights for basically free as well and be competitive with the ground based systems.

Trying to get the real table to update the ball at 20fps is proving insanely difficult.

Maybe on a 20Mhz 386, but by the time Windows XP was being tested, it would do over million FPS so the XP version was hard-limited to 120FPS, dropping the CPU to under 1%.

https://devblogs.microsoft.com...

Age-verification at OS levels was always a terrible idea. It's difficult to see under what rationale Linux should be granted an exception for this dumb idea. The solution is just to repeal the law and flog the sponsors.

Well, the problem is age verification to begin with. But since we have some states wanting age verification, it's a privacy nightmare. OS based age verification seems to solve the largest problem of all - needing to submit to a third party your ID information to confirm your age. Because they've all been hacked and that ID leaked But if you can embed that in the OS - and Apple and Microsoft know information about you to get a good estimate (e.g., if you have a credit card), then it can be used to attest your age instead of having to submit to the third party.

That's why they don't mind doing the Linux exception - because if the OS doesn't want to do it, then those users can rely on the third party services. Those third party services will exist anyways to handle the many OSes that won't work.

If you think the CloudFlare interrogation is bad now when you use a VPN...

Probably it will be a posthumous Darwin award.

I don't think there have been any Darwin awards that were not awarded posthumously. The award requires you to have removed yourself from the gene pool in order to be considered for the award.

I do love it when malware advert javascripts can upload random new firmware updates into my mouse and keyboard turning them into stealth keyloggers. This is great.

This feels like when Flash sandbox breaks became a thing, but worse. At least in those days we got smooth fullscreen vector animations and games to enjoy. I'd rather Flash had just been bloody fixed instead of browsers themselves becoming Shit Flash But Holy Cow It Runs Worse And Gets Worse.

You'd have to really be terrible to let it happen. First, you have to authorize the device to be accessed - and almost always web serial devices are using libusb. They have to as no OS allows direct access to USB devices - you must always go through a driver. Libusb is the only thing that really pipes a USB device through to userspace. And if you're using libusb, the OS driver is not running.

And to accomplish this, you almost always have to override the OS settings to prevent loading the OS driver over libusb, especially for things like keyboards and mice. It's possible, but it's complex, and it's why in the early days, you had many peripherals saying "do not plug in without installing driver".

Honestly, it's far easier to develop just a malware program in general than to try to break out via web serial. And if you already have the user to run the malware, why bother with web serial at all?

Also, it's a permission you need to give a website, and almost none request it because it's only for web=based IDEs to program embedded things.

You want a larger surface area, you attack things like WebGL, which you'd want to do as there are performance critical paths in getting from the browser to the GPU, and many of those paths are not protected very well

Then again, how many projects use Xilinx parts?

Most FPGA projects are using Intel/Altera parts - MiSTer, etc.

I can't understand the thought process behind them making everything public by default. Why on earth would anyone want personal financial transactions public?

That's the first setting I changed when I installed the app. I don't use it much, but some people prefer to be paid that way.

Likely it was a "social" thing. They want you to use them for things you do in public - I sent you my share of the restaurant bill and that becomes part of the social media chain of events - you went to a restaurant, and then you paid your share via Venmo to your friend.

It was supposed to be something you did like you posting food on Instagram and other things.

The only problems came about when people started using it as an alternative to Paypal (even though it was owned by Paypal) and having those defaults meant some rather shady transactions became exposed. I still don't really get why you'd Venmo people instead of Paypal them. It's the same company.

I was thinking the same thing on the average. I'm assuming executives will get the biggest share of that, while the common workers will get, what, say 10k? Not that that's anything to sneeze at, but let's not pretend the average low-level person is getting 340k.

No, it's likely the union members get a big chunk of it.

The reason they went on strike is SK Hynix (also Korean) gave the excess profits to their workers - to the tune of about half a million dollars each. Samsung executives refused to do same when Samsung employees asked so they were threatening to strike over it.

So it's likely going to go to the workers. The executives were going to get their cut anyways.

Is there a guideline somewhere on how these groups run fully anonymous infra?
Do they get IP blocks lent by someone? Wouldn't they need a real name and card to pay for hosting, DNS, etc?

It's less about the repository and more about the whole lifecycle tracking it provides.

It hosts code, but it also provides an environment to report issues, and have those issues tracked through to a commit. It also makes it easy to handle contributions from other people.

If you have a small project, it's no big deal to do it manually. But once a project gets big enough, you really want something to track bugs, support tickets, fixes, releases, pull requests and code reviews.

And "big enough" is basically at the point where you need to have more people.

GitHub, GitLab, etc., are going to do fine - because the main customers are corporations who run their internal self-hosted version (enterprise editions) because it's the exact same issue managing 100+ developers on a codebase.

Linux is special - and they had a whole development methodology set up for large scale distributed development long before things like SourceForge even existed.

Well, to be fair, the memory companies have suffered the past 3-4 times that memory prices skyrocketed, they brought more factories online, and then prices plummet just before the factory comes online so they're selling increased capacity into a surplus market.

It's why a company like Kingston exists - Kingston exists solely to absorb surplus memory. If memory makers make too much RAM, they sell it to Kingston and Kingston makes a bunch of memory modules from it. Likewise if they make too much NAND flash, it goes into USB sticks and SSDs. It's why Kingston RAM and flash products are so variable - they just take the surplus parts and put them into products. You might get Samsung products one day, the next day it's SK Hynix because that's what's coming in the factory.

None of the memory makers are bringing up their timelines of increased RAM production from the late 2020s/2030 because they don't want o bring it online into a market th's flooded.