Airplane mode
On all recent-ish phones you can turn on airplane mode then re-enable Bluetooth and/or WiFi, leaving only the cellular radio disabled.
7839 publicly visible posts • joined 16 Jun 2009
Walkie talkies will even trip most RCD/GFD and AFCI/AFDD circuit breakers if "clicked" within a few cm of them. In some cases they'll even trip MCBs.
They have an absolutely massive output, as that initial burst is basically unregulated.
Never use your radio near electronics.
It's one reason the emergency services have been changing over to "mobile phone" technology. The other being encryption, of course. "Police scanners" don't pick up LTE based trunk calls.
When you wander into an office building on your phone it's going to lose the LTE signal and pick up the WiFi. Under TCP the connection to the remote server is broken because your IP and port changed from IPv6 on LTE to a NAT'd (possibly CGNAT'd) IPv4.
Session IDs let the user roam from LTE to 3G to WiFi to wired without the application layer having to handle a reconnect and resume. In theory it even allows applications to trunk connections.
So that long-running download doesn't have to be restarted, only a few tens of lost packets resent.
Most laptops too. Windows and macOS do this by default to all WiFi adapters now.
It's a pain, as a lot of single-seat software still ties the licence to a randomly-selected MAC address. When that's the WiFi, you end up having to move the licence to a "new" machine every single day...
For there to be 13% who know someone who has.
On the other open claw, nobody is going to admit they sold it - even if they did. So the question is probably the best approximation available.
The C-suite figures are both horrifying and as expected, mind. Many of them intend to have moved on to another company before the fecal fan interface occurs.
In the real world there always is a central, golden repository.
For example, the one that Linus Torvalds pushes to.
The genius of distributed version control is that everyone who wants to can have a copy of that golden one. That means it can vanish and a new one set up, and all it costs is a few seconds pointing your local copy towards the new shiny. No data will be lost.
This is of course why GitHub et al added those extra things around git - the PR and issues discussions, wiki, Actions etc. Those are the things that keep projects trapped in fluffy handcuffs.
For how long though?
The Github $39pcm plan would have cost £1000 in about 34 months. So including electricity and assuming a five year replacement cycle means the local hardware is pretty similar cost.
But the $39 per month plan is gone, replaced by something roughly double the price. Now the local hardware is cheaper after two to three years.
Yet Microsoft are still losing money at the new price. The cost per token is going to rise further. In a year that $39 pcm usage is going to cost much closer to $400 pcm, and your local hardware will be cheaper within six months.
The reason analogue clocks are ending is precision.
We used to only really care about the nearest hour. The single pointer on a sundial or similar was plenty.
Then we started caring about the nearest quarter hour, and needed a second pointer. That worked really well.
Now my train leaves at 17:38.
It's very difficult to see arbitrary times on a distant analogue clock with better precision than about ±3 minutes.
Even if the marking are clear, parallax and mechanical wobble reduces the possible display precision of the moving pointer. It's no good if I see the minute hand pointing at 17:36 if the dispatcher along the platform sees 17:38.
Who decides what's "correct and complete"?
When the DVLA lose the paperwork, how do you prove it?
And if the DVLA claim they asked for more information but didn't actually bother sending the letter - or Royal Mail left it on a shelf for six weeks.
Cleaning up the mess when the DVLA screw up is difficult, because everyone relies on their database. Buttle/Tuttle is just the start.
Several friends have had similar incidents with visas. HMVI lose half the paperwork and the visa gets denied "due to lack of evidence". Despite what the actual law says, there's effectively no recourse whatsoever.
The MS account requirement utterly screws small businesses who want MS365.
Tiny business buys a laptop. During first boot they're forced to create an MS account without realising it.
Once they have a PC, they buy an MS365 subscription tied to their business email because they think they want Word and Excel.
Now they have two MS accounts on the machine and no idea how to get rid of the one they were forced to make during first boot. Their experience of MS365 becomes orders of magnitude more horrific than normal as they have to keep switching between those two accounts all the time. MS365 also keeps insisting that it needs to take over their email hosting, and if they follow that "guidance" it won't work...
Worse, they have no idea that bitlocker was enabled until a bad update requires them to find the recovery key. They now have absolutely no idea which account contains said key.
Or, they use LibreOffice and have none of the above issues.
It seems the original contract assumed the suit manufacturer would liase with the lander manufacturers to write the specification for suit/craft interactions.
Predictably, that doesn't appear to have happened. Apparently there's still no specification for what needs to be in the donning/doffing chamber - or even how big it needs to be! This makes it somewhat difficult for BO and SpaceX to even begin designing the lander interiors.
Standards matter - it's why Dragon can dock with the ISS.
Companies often pay for things that aren't useful as an experiment to find out whether it might be.
On top of that, if something is useful at $10 or $39 per month, it does not follow that anyone will pay $500 or $1000 per month for it - and that's the kind of price rise that must come.
Residential routers are on 24x7 - as are commercial ones.
A single residential router is worth nothing to a miscreant, but a bot net of many thousands of them is very valuable both for DDoS attacks and for attempting to break into something that is valuable. They all come with their own IP, so such things come at you from all sides.
Most people use the router their ISP gave them and rely entirely on that ISP to keep it secure - after all, most people don't know how any of this works and shouldn't need to!
A single serious vulnerability in one of those routers can give an attacker a significant proportion of an ISP's customer base.
I have to fill out a tax return every year.
So I need some method of logging into the HMRC systems to send them my completed tax return and request my refund.
There's also various benefits that most people can claim some of.
How else would I do this?
Of course, none of these required a common "Digital ID", and would in fact be considerably less secure if they did.
Onboarding is an absolute nightmare across the board. Almost everything is actively user-hostile, and the only reason anyone does it at all is because they don't have a choice.
The only exception seems to be TOTP on a smartphone via QR code, but that still means installing a TOTP app - and figuring out how to have a backup.
All the code review LLMs moan about variable names, they've had that feature for ages!
The code being scanned has been sent to Anthropic, and they promise* they won't keep it and use it for training.
I've found these code review tools to be a useful first pass - that regularly flips to being an infuriating waste of time.
At the moment I think they're probably worth the price. However, I know that they're being sold well below cost and the break-even price is almost certainly going to be way out of budget.
So I'm never going to put them in CI, and I don't want my team to become reliant on them either.
*How would we ever know? Our stolen closed-source code would get used by other Anthropic customers for their closed-source code. Best case is it turns up in the discovery phase of an entirely different lawsuit, long after Anthropic have gone bust.
There are economies of scale, and smaller organisations can't afford the expertise needed.
Local government like small town and parish councils often have one part-time employee, and around 80-90% of the budget is spent on their salary.
There's no way they have the expertise or even the time to look after a single server in a colo, yet this organisation is Government and does have a fair amount of privileged information to protect.
So they must come together and share the cost with many other similar organisations.
At the moment they're all being pushed hard towards MS365 (or sometimes Google Workspace). Those are very expensive (well over £1500 a year) and hand everything to the US Government under the CLOUD Act - so possibly unlawful, but it's unclear. MS365 is difficult to manage, so they have to hire a company to do that too - which means many of those are pushing very hard to resell MS365 and get that juicy password reset money.
Having multiple sovereign options "blessed" by the EU Commission would make it far easier to avoid that trap.
Not even that, because it doesn't tell you anything about the gaps and thus very little about the direction of travel.
Are Oracle within a couple of percentage points of the one above - or the one below? Are they rising towards the one above or falling away from it?
This kind of chart is light on information.
A device whose screen swivels around to face the person speaking during a group video call.
Wow. If the best innovation they can publicly mention is to tape an iPad to the top of a 2017 Meeting Owl, then the new guy is the same as the old. Surely it would have been better to say nothing?
I guess business meeting hardware would be a new market for them, but they've historically avoided making any explicitly business oriented products as that might affect their "premium mass-market consumer" reputation.
Unlike the Neo, as extending a little downwards in price is usually a very good idea.
There's a lot of NTFS formatted partitions out there.
Most of them are in PCs running Windows, where the owner has yet to be convinced that they could switch to Linux.
They're not going to want to lose their existing data or spend a lot of time restoring backups onto a different filesystem.
The fewer barriers, the more likely they are to consider trying it.
The usual reset logic seems to be a timer.
Starts in state 0, first button press starts the timer. Any wrong button press puts it into state -1 where all buttons loop back to -1. Timer expiry resets it to state 0.
The final correct button places it into state Unlocked and stops the timer.
Some have an explicit "reset" button as well.
Usually the keypad is the most secure part, of course. It's the rest of the lock that's useless.
It depends how soon they crash.
If they crash out this year, then the small to mid-sized useful models, especially the domain-specific ones, will remain available for individual businesses and researchers to run and retrain on-prem or on hired servers.
If it takes much longer than that, the loans will be so huge that we're in 2008 financial crisis zone and the real economy tanks along with them due to banks going bust.
In the latter case, investors will be so spooked that only the models that can be usefully retrained on desktop workstation hardware will remain in existence.
So for the industry to survive, it must die very quickly.
In most industries we kill a product or even entire brand once it becomes clear or even likely that the minimum profitable price (possibly bundled with something else) is too high for enough customers to actually pay.
It's been blantantly obvious for a very long time that they need to add at least one if not two zeroes to their prices to merely cover their borrowing costs, let alone ongoing running costs.
There's no way they can make a profit without bankrupting nearly all their customers, and that's not even a short-term plan.
Mobile phone networks worldwide are mostly v6 only.
The UK and USA wired ISPs are mostly v4 only and using CGNAT, because it was cheaper than upgrading their existing "free" router estate. (Which they're now having to upgrade anyway. Except in the US, where it's now illegal to upgrade routers)
In the rest of the world, older ISPs tend to be v4 only while smaller ISPs are v6 only.
Virtual servers from the "big cloud" providers are mostly v6 only with a translation front end, because v4 is expensive. Colo/on-prem seems to be a mix of v4 and dual-stack.
ISPs and cloud providers are translating in both directions.
We're still in the transition period, and will be for at least another decade.
The committee did.
In fact, most of the complexity came from the fact they did listen to that feedback and put in ways to do it. Translation, autoconf, etc. Probably too many options.
The only reason IPv4 still exists on the global Internet is that ISPs decided to CGNAT your NAT instead of using the much simpler v4 to v6 address translation that's in the standard.
I have no idea why they went that way, as CGNAT is way more complicated and has a lot of unwanted sideeffects.
But then, I'm still seeing customers insisting on static IPv4 addresses and cursing that they need to manually set a number in everything. They still refuse to DHCP because "they need to know the IP". Yet they absolutely don't, it just causes them pain.
After a power-cycle or firmware update, you must unlock using the onscreen pin or password before any external devices will connect.
I don't think FaceID or fingerprint reader work either until you've unlocked it once. (Certainly this is the case on Android)
I am wondering how the OP is going to install this update, given that it cannot be unlocked. Maybe there is a force firmware update if booted in the recovery/wipe mode.
The GPUs need CPUs with lots of RAM, the boxes need to go in a rack somewhere, they need huge amounts of power and data interconnect, and huge amounts of non-volatile storage.
And all that consumes unimaginably huge amounts of money.
Which means everything else falls by the wayside.
Surely all you need is to include previous cost overrun and underdelivery weightings.
If previous similar contracts had a 500% cost overrun, their bid price is multiplied before comparison.
If previous contracts were underdelivered, it is reasonable to expect this one will be as well.
Procurement rules already include risk weightings for these, so use them!
The FAA generally requires first-time applicants to be under 31
Well, there's a major part of the problem.
In 2025 only 13% of the US population fell into the 20-30 age group, so their pool is really quite small.
Worse, very few people can afford to burn the waiting time at the start of their careers, as they don't have the buffer from existing or previous jobs to tide them over.
If I go with VM, (shudder), has anyone successfully escaped from a contract with them citing inadequate service?
Yes, but it wasn't easy and relied on consumer protection law. It should be much easier now though.
Except you're a landlord, and thus may not get consumer protection at all and may even be forced onto a Vermin Business plan.
This FCC router policy is indeed banning "foreign assembled".
However, there is an explicit statement saying they might let you import if you promise to start manufacturing them in the USA "soon". (And an implication that greasing appropriate palms might also work.)
Worse, even existing hardware only gets a pass for a short time (I think two years, might be three).
It's why the industry has started screaming, because right now there are (counts on fingers) none.