The difference is because Unix has two-phase deletes, while Windows doesn't, inheriting the delete semantics from DOS and a bit from VMS.

Two-phase delete refers to the act of deleting a file - if you delete a file on Unix, the file gets marked as deleted, but the actual file blocks are not deallocated until the last reference to the file is closed, at which point the file blocks are reclaimed. This can leave orphan file nodes if a delete was scheduled but not done (e.g., power loss, sudden reboot without unmount, etc). This is why you have a "lost+found" directory on Linux filesystems - file blocks that were deleted but not actually reclaimed end up there to indicate the directory entry was removed, but the inode was not reclaimed.

On DOS, deletes are blocked if a file is in use. For a single tasking OS, this is perfectly normal as the last reference to the file should be gone by the time you delete the file. Of course, with the introduction of multitasking, and networking, these things become more complex. (MS-DOS's SHARE.EXE was a program used ot manage shared files). This leads to issues when deleting a file since you cannot delete a file that's in-use.

This is also why there are tricks on Unix where you create a file, then delete it immediately and you then use it as a temporary scratch file that will go away when you're done with it without littering the filesystem with random files. You can't do that on Windows.

When you do an update, you may be replacing files that are in-use, e.g., system libraries. On Unix, you can do this - programs already running continue to use the old library on disk whose reference was deleted, while new programs will use the updated library. However this can lead to issues since now you can have programs with a library mismatch - one program is using an old version of the library, while another program is using the newer version, so you need to restart the programs using it. And sometimes this can lead to bigger problems like file corruption if the file format changed and programs are using a mix of the and new libraries.

So in general, a reboot is generally safest unless you are sure no one was using the file. Windows needs it more as it can't update files in use so the early boot process the system performs the updates and deletes that were scheduled, while Unix they may reboot just to ensure everyone is using the right version of a library.

If Tim Sweeney wants full openness from app stores, then all stores should be o pen. Including the Fortnite store. I should be able to get dances from other stores if I want.

Given the money Fortnite makes for Epic and Tim, it should be fair play to demand the same. Fortnite is like the #1 game after all.

Easy. If you use a ticket, it's not tied to anything other than say, a credit card. If you sell the ticket, that ticket needs to be presented with the original credit card used to make the purchase.

If you sell your friend the ticket because you can't go, chances are you trust them enough not to abuse your credit card that you can lend it to them. And if they do go wild, well,they're not your friends now.

This has been used before and was the basis of some of the more creative DRM systems out there - where the key to unlock it was based on the item ID and credit card number used to purchase it. You could easily generate the key knowing the credit card number, but that also means if you want to spread it around you have to give the credit card number to everyone.

And another solution is simple - all tickets are refundable - if you find you can't go, you can return them and get your money back. If you can't get the tickets, you can check back often to get tickets which might have been returned and buy them.

"Flip phone" in this article is referring to foldable touch screen smart phones, unlike every other use of the term.

Yeah, the article talks about what we call "foldable phones" and not "flip phones". The former are phones that unfold and reveal a huge one piece screen, the latter are things that open Star Trek communicator like where the screen is on one side and the keypad on the other side.

Or has Gen Z completely decided that folding phones are now flip phones?

The problem with Chinese EVs is first partly because they invested so heavily in R&D, they ended up as the leader in EV technology.

CATL is one of the largest battery markers and is basically leaps and bounds ahead of the curve on many technologies including LiFePO4 batteries as well as sodium-ion. If you want LiFePO4, you have to go CATL, and sodium ion is pretty much exclusively theirs.

It's a reason why BYD can make a $10,000 EV that's actually decent. Bit spartan but for the price, you get plenty.

The Big 3 are unfortunately, completely uncompetitive companies. The only reason they're around is because the domestic market is pretty much their lifeline - being the only market that wants the really big vehicles they make. And they went to bigger and bigger vehicles with clever marketing since the 90s - a bigger vehicle not only doesn't have obey CAFE, but they make more profit for the companies involved than a smaller vehicle. They're so addicted to the profit that they cannot build a smaller vehicle anymore (at least, one with reliability). Pretty much ceded that market to the Japanese and Koreans.

(And Trump expects Japanese consumers to buy the F-150 in quantity when they'd much prefer their Kei trucks that are half the size and more nimble to navigate through the congested roads of Tokyo). Meanwhile American consumers who want small efficient cars have to buy Japanese ones.

It's probably time to admit that the Chinese are really good at EVs and engage in technology transfers to bring some of that over - get BYD to build a manufacturing plant and produce their cheap cars. Sure we can say "no cheap Chinese labor" but even so, I'm pretty sure they can build an EV with American labor for under $20,000.

Maybe you're planning on staying overseas for a length of time and want to learn the language to communicate better?

If you're a tourist, a translation app might get you by. But if you're planning on a long term stay in say, France, learning a little French might not be a bad idea.

Limiting yourself to places that only speak English is somewhat limiting. And while English might be the language of business, plenty of business is conducted in other languages. It's not unheard of for suppliers to cheat Americans by adding a little "doesn't speak local language tax" to the price.

Or maybe you're a top tier researcher in your field, and your services are requested at high cost. It might help to learn the local language so you're more equipped to handle the local language technical jargon of your field.

Or, until the ICE crackdowns, you wanted to tell your Spanish speaking workers how to do something, and to be able to hear if they backtalk you

Lots of reasons to learn a new language. Whether Duo Linguo is good or not, who knows. There have been many other learning packages out there.

And the problem is ChatGPT could basically duplicate what Duo Linguo was doing by helping people with basic foreign language, so it hurts their business prospects because you could get for free what they would show ads for and make money.

Even fairly middling mail filters get a lot of the really lazy stuff; and if you don't want people clicking on Important.doc.exe you just tell the mailserver not to give it to them; not try to train them out of double extensions. If they keep falling for fake login pages; well, that's what the FIDO2 requirement is for.

That filter doesn't work because many phishing scams use encrypted files, often with the premise of "protecting your information".

Of course, the password to use is given in the email. so higher end mail filters do scan for the password to try to scan the archives, but using encrypted archives is a great way to bypass mail filters.

It's open source. If a company wants a feature because it helps them, they have two options - do it themselves (and either release the change or keep it to themselves), or ask someone to do it.

Asking the maintainer for a feature request happens all the time - most software projects have feature requests lists a mile long. It's up to the maintainer to pick and choose what they want done - maybe a Fortune 20 company makes a feature request that sounds fun to the maintainer - should the fact that company never made any contributions be a consideration?

If the maintainer wants to work on it, they can work on it. If someone else finds a feature fun to implement, they can work on it and contribute it back.

Of course, such feature requests are often done on an voluntary basis - it'll get done when it gets done. Maybe an individual request is so boring no one wants to do it - to deprioritize a company's request which may be more "fun" to do would be silly.

Open source software should be considered "I did this because it was fun, if it's useful to you, great!". If a company makes a request that's interesting, it would get worked on. If it's a boring request, it'll get ignored.

If a company really really really wants a feature, then they're free to make inquiries into hiring someone to implement the feature, or paying the project to implement the feature. That changes the operation from one working on it for fun to maybe working on it to get a few dollars. Which is also fine. And the maintainer is free to not take the money as well.

I see nothing wrong with freeloading companies - if they find my software useful, great, if not, oh well. If they find my software really useful that they package it with other software, also great. They're also free to make requests just as I am free to not honor them, or ignore those requests.

If you're worried about people making money off your code, maybe open-source isn't for you. There are plenty of other models you can use that may suit your sensibilities.

90% efficient panels? Photosynthesis is highly inefficient.

To convert to energy it's typically under 6%.

Nearly half the light falls outside the visible spectrum, which isn't used by plants (photosynthesis only works between 400-700nm). And those higher energy photons must be converted to lower energy photons as the actual conversion requires 700nm photons. And there's a good chance the photons hit the wrong part of the leaf, so by the time you get to the actual conversion you've lost 70% of the light energy. Another 66% is lost converting that to sugar and energy products, so 10% of the light energy actually makes it.

At night, that sugar is consumed to continue life processes (which also means the leaf consumes oxygen) so at the end the leaf has around 6% leftover. And this respiration by trees is why their actual oxygen contribution is far lower than they're given credit for.

Meanwhile, commercially available solar panels are achieving 15% with lab demonstrated 20+% efficiencies.

Trees haven't had a need to evolve more efficient photosynthesis since there's just a lot of sunlight available.

Yes and it's similar here. You up the performance, you lower the efficiency. Selling it after market helps them keep regulators happy with good numbers at the time of sale, but gives consumers the option to have more power with lower range. Not defending it, just part of the game unfortunately.

Presumably the tuning also voids the powertrain warranty, so I wonder if the subscription price includes warranty coverage for the tuning you get.

This could be worthwhile simply to not void the warranty on your engine and transmission.

SGX is also deprecated. Support ended on the Gen 12 CPUs, but later on Intel disabled it completely on Gen 11 and Gen 12 CPUs so it only exists for Gen 8 through Gen 10.

There is no current Intel CPU supporting SGX.

Who is on their board advising the CEO to ignore Apple in wearables? Another strange item I found was that the CEO was recently ousted for investing decision of one billion for a head phone company (Bower Wilkins)?? What was that person thinking? "F apple, I'm getting into the headphone game now, I'll show them?"

Likely because they're a medical device company and Apple is a consumer electronics company - and Apple wouldn't bother going through all the hoops the FDA has to make the Apple Watch approved for medical use... would it?

Medical device companies exist in another world - where they can build a hearing aid that costs $5000 for a pair, and then charge you $500 if you want Bluetooth with it (via an external box).

Of course, the electronics are fairly cheap - which is why you can buy "hearing enhancers" for $150.

Or, a pair of Apple AIrPods for $300 which are now hearing aid certified - which support Bluetooth as well.

They got the claims dismissed after all.

That just made the legal suit go away. They still incurred costs that they didn't get reimbursed for - like say, their lawyers. They had to deal with the courts, a lawyer had to draft a response and deal with all the paperwork.

And this dragged on for 3 years. So that's 3 years worth of some lawyer's labor in handling the issue.

So just because it got dismissed means that everyone is right back where they started - I doubt Visa got reimbursed for their expenses and lawyer costs (even if it was a corporate lawyer working for them), so they're still a net negative in the end. In other words, it cost them money, and companies hate it when that happens; So they're going to avoid doing activities that repeat themselves like that in the future. Usually this is done by having adult services pay more in fees to overcome the higher costs they generally incur.

As for Steam, well, chances are they didn't want the risk. But they could be persuaded if Valve either indemnifies them, or pays the higher fees associated with the higher risks (including risks of a lawsuit).

Or maybe there isn't one and someone will step up. There probably will be several successors, and the one that that will be "it" will be determined by who the developers are more willing to work with.

The lieutenants do a basic job of maintaining their subsystems and chances are there is probably someone they feel can step up into the role. If there's a fight, it would be fought by whom the developers pick to work with.

It's an open source project, and the Linux tree is the official tree, but there are trees that other people maintain and they pull patches that aren't in the Linus tree all the time.

If there are multiple trees, then someone will pick patches from each that make sense and now there will be a new tree. Over the course of months eventually it'll get settled by what users and developers choose through their actions.

It's open source. Forking is expected. If there's a fight, then people will pick a fork.

What I don't understand is why there is no card/slot for M.2 drives. Even a full length one is not much larger than a USB drive, and the slot could be made similar to a larger USB-C style connector. Is there something technological, like trace length limitations, that makes that difficult?

There is.

It's called SD Card Express - and yes, we're talking the same as the Nintendo Switch 2, which was one of the first widely available devices to use it. It's basically NVMe in an SD card form factor.

There's also CFexpress which uses NVMe.

If you want PCIe, we call that Thunderbolt.

Of course, to call a card slot M.2 is confusing because M.2 could have USB, PCIe and SATA connections on it, all of which have external counterparts. Heck there was an oddball USB/eSATA combo port for a while as well.