Posts by bazza

Atlassian migrated 4 million Postgres databases to shrink AWS bill

Re: Foolish nincampoop

Mulling on this throughout the day, I've developed the thought further.

Suppose AWS's devs did use Atlassian, and considered its "vital" to their work. If Atlassian plonking their tool suite wholesale on to AWS got thought about, AWS's devs should by now be thinking, "We've got to find an alternative". So AWS sets its software dev teams building or improving an alternative. The latter is more likely, you always want an easy start, and it's probably OSS.

So AWS works on it, produces an OSS equivalent of Atlassian's tools and keeps it OSS, and Atlassian's market evaporates all of a sudden.

At the same time, AWS should recognise that destroying the market for one of its key customers is not good for its own business. Atlassian goes bust, that's cloud rental income lost.

So is there some sort of tricky issue here for AWS devs (and devs on other clouds too)? Sure, there's a point where if a cloud operator doesn't trust its own cloud services then why should we. On the other hand, with a lot of current wisdom being that it's good to keep at least some stuff on-prem for "in case", how does a cloud operator get some on-prem resources of its own. Or rather, as a cloud operator is all on-prem, from where do they get some independent resiliency? Another cloud operator?!

Re: Foolish nincampoop

Also, we better hope AWS own dev team don’t use Atlassian. Imagine:

Bezos: “AWS is down, fix it”

Team: “alright we’ll look at the Jira tickets”.

Bezos: “And?”

Team: “We don’t know, Jira is down.”

Re: Atlassian

It does sound like they’ve done “close down one instance of Postgress and start up another pointed at the same db.

I guess that keeping the service up whilst doing this could be tricky?

Also, watch Amazon change the pricing structure underneath them. Atlassian is saying “we’re making more money”. It’s only a matter of time before Bezos says that some of that is going to go into his pockets.

The year of the European Union Linux desktop may finally arrive

Re: Desktop is not the only problem

Catia moved to Windows only a long time ago. Lots of these former Solaris/Unix CAD applications have.

Re: Desktop is not the only problem

Rewriting on other platforms could be seriously tricky. MS has made Windows a really good platform for very heavy duty 3D graphics. Games benefit from this, which is why there's a lot of games that are on Windows and not on Mac (or Linux). Porting some heavy duty applications (on the scale of Catia) to non-Windows could be a serious downgrade.

I mention Catia because it's French.

French city of Lyon ditching Microsoft for open source office and collab tools

Re: Hardly a surprise

Forking open source is a risky business. Unless you bring a big chunk of the developers with you, your fork likely dies.

And these day RedHat has some key projects under its exclusive control by employing the bulk of the developers.

Amazon's Ring can now use AI to 'learn the routines of your residence'

It Won't Do Them Any Good

The whole Amazon Alexa / Ring IoT thing is famously losing Amazon a lot of money. I struggle to see how adding this kind of thing in is going to make a difference.

Google has had Nest thermostats in millions of homes globally all with the capability to determine the rhythms and patterns of life in the household. Yet, they're abandoning Nest thermostats across a large portion of the world. It's evidently not worth their while. I'm not sure why it'd be worth Amazon's while.

The cynical part of me is that things like Alexa / Ring continue live whilst the developers can say "hang on, we've a great idea that'll help turn things round". As soon as they say they're done, they've developed these things as far as their imaginations can get them, if Amazon are still losing billions on the business they're going to close it down and those developers are out of a job.

Japan's sequel to Fugaku supercomputer will be Arm'd to the teeth

Re: Predicatble but cool

It’s going to be pretty good in terms of peer per Watt!

This, Fugaku and the K Machine are the last of the great supers, with bespoke hardware (specifically, their Tofu interconnect). The others are now just stuff like what we mere mortals can buy but piled up in large quantities.

LibreOffice adds voice to 'ditch Windows for Linux' campaign

Re: Problems

Of course I realise all that. I'm simply pointing out that none of that serves to make Libre Office (specifically, the whole end-user experience of using Libre Office on a Linux) desirable to the majority average computer use. And it's not the fault of Libre Office.

Libre Office can become more popular on Windows and Mac through publicity. It's not going to get substantially increased numbers of Linux users via publicity. It's already got 100% share of the Linux market (effectively - anyone using abiword?), so it needs Linux's user base to expand. But that seems most unlikely.

Re: Problems

600k+ application isn't the problem. Linux adoption and dominance has not happened despite a vast library of software, curated or otherwise.

One aspect of the problem is the very fact that the packages have to be curated in the first place. Linux suffers dreadfully from there being far too many ways to distribute software for it, hence the need for curation. "Stray off that curated reservation at your peril" is the message for the lay user. Most other OSes - including other major free ones - get software installation right.

Also I don't think you quite appreciate the level of integration between apps, Windows and AD. All of Edge, Chrome and Firefox can be controlled by GPOs in Active Directory, provided they're running on Windows. Linux, not so much. Linux achieves no such level of integration, neither with AD or a OSS equivalent, even if the component parts might theoretically be lying around the place. In fact at least two of these try to make the admin's life hard by offering things like DNS over https.

It is this kind of missing functionality plus all the other hoary well known problems in the Linux world that discourages Linux adoption on the desktop / laptop in a corporate environment. That is why few choose to use it at home, so few go on to pick Libre Office. Libre Office itself on Windows makes an awful lot of sense. Unfortunately, so too does MS Office.

Re: Problems

And on how many of those 4,500 Linux systems are there users using Libre Office? And of those, how many are long-experienced Linux admin types and how many are your average non-IT corporate employee?

Re: Problems

Such as the Danish government's department for digital affairs. I am wondering if it's just coincidence that Libre Office's blog post came out 2 days after that Danish department announced their intention to deMSify themselves. There's no problem with that of course - well timed / coordinated publicity is good publicity.

Libre Office itself is perfectly capable. I use it. I even deploy it using Windows Group Policy (though their MSI's have been a bit broken in that regard of late). It even ticks the box for "not mucking around with the user interface too much and making user's life hard". The biggest challenge to its wider adoption isn't what Libre Office itself can do. It's that for those intent on running it on Linux in anything like a corporate environment for those users not thoroughly attuned to using and maintaining Linux (i.e. the majority), life can get hard, fast. And that if your org rolls Windows machines to solve those problems, MS Office is a very natural add-on.

In the Linux world, the app stack and OS are separately governed (and probably never talk). MS's advantage is that they're doing both the OS and the app stack. The difference shows up in adoption rates.

Problems

Accessibility.

Platform management; say what you like about MS, but device fleet administration is a whole lot more sorted out on Windows than anything else. There's a reason why corps by fleets of Windows machines.

Training.

Support, especially hardware support

PCIe 7.0 specs finalized at 512 GBps bandwidth, PCIe 8.0 in the pipeline

Re: Mind blowing

Well, technically, at that speed it's not down the copper track on a motherboard. The energy is actually RF EM fields in the dialectric between the copper track and a ground plane (i.e. another copper layer). So the material which the board is made of is going to be important.

It's been like this for a while. The tracks between memory SIMMs and the CPU have also effectively been RF transmission lines for quite while, which is why they can have different data bits on them at different points along the track.

Ethernet is another.

These days high speed digital data transmission has far more in common with radio design than it has with logic design. Ethernet doesn't have 1's or 0' on the wire, it's a complicated some-sort-of-QPAM analogue signal. Put it through an amplifier and aerial send it through the air! Only at the end points does it get turned into 1's and 0's. I suspect that to get across a PCIe bus, it's also transformed back into a similar analogue transmission.

Board manufacture is already quite difficult, and it's going to get more difficult. Track impedance control becomes more key, the faster the signalling rate. Printed circuit board technology is pretty good, but it's always been a bit tricky to control track impedance.

There have been alternatives. Multiwire is another - think of a pen plotter drawing enamelled copper wire on to an un-set layer of epoxy, drawing little spirals where there needs to be a via, drawing wires crossing over other wires. The Japanese still have this capability I think but it disappeared here in the West a long time ago. That board technology had really good imepedance control, largely due to the very good control one had over height of wire above the ground plane, and dimensions of the wire itself. I wonder if this will make a comeback?

Interestingly, if one wanted an optical circuit board, that's probably how one would make it; a Multiwire process, but drawing fibre-optic out into the epoxy instead of enamelled wire.

Well, that'd probably suit my needs. Just about.

The idea that PCIe8 may be optical (even if optionally so) is rather remarkable. There's going to have to be some neat developments in connector technology! Or is a PCIe8-O peripheral simply a cuboid with a couple of ST connectors on it and a power connector?

At some point I suspect Ethernet and PCIe will merge. Everything will have a MAC addr... That could solve the problem of Wayland being naff from a remote desktop point of view. Instead of all that complex piping of compressed framebuffers, simply have the remote app attach to your laptop's GPU over the network!

Floppy disks and paper strips lurk behind US air traffic control

I gather it now costs comparatively little to get obsolete sillicon chips remanufactured. There's company(ies) that have specialised in preserving the lithography masks, and what was once a costly shit-hot 100nm fab is now a not very busy 100nm fab that's open to any work going.

I hear that this has had an impact in some military systems upgrades. Apparenty it can now be cheaper to get old electronics re-manufactured and brought back out of the "obsolete" bin to run existing binaries than it is to stand up a software dev team simply to port the existing source code to a newer version of Linux (a job that should be no more than ./configure, make, make install).

Ok, this is not exactly "routine", and it's full of ifs and buts, but the fact that it is happening at all is remarkable. This is especially remarkable as the whole reason we have portable source code at all is partly thanks to the DoD mandating POSIX (a standard that kinda coallesced what we think of now as Unix, and prevented more OS fragmentation than already existed at the time). Except now the software compiling is the expensive bit.

This ought to be a wake up call for the software industry. It's long been the case that the myriad minor variations in software development ecosystems has inflated the costs of keeping software alive. Variety is good, but it's also bad. Every time someone decides to tweak how cmake works, it costs someone somewhere a fortune in updating their build environment (or creating a fresh build environment). These day, for anything even slightly chunky you now need a devops team simply to manage the build environment, on top of the engineering team actually writing the software. This is going to kill off software development. As tools / applications mature and become "just fine for the job", the shop who can keep that software running as built 20 years ago is increasingly going to be the winner. Cloud won't help because all Cloud does is mean "at least you're not having to maintain hardware".

Trump lifts US supersonic flight ban, says he's 'Making Aviation Great Again'

Re: Not those Boomers.

Quite a lot of Concorde's regulars were businessfolk in New York who'd use it to hop across the Atlantic for a meeting. The events of 9-11 unfortunately killed off a large percentage of the regular customer base who'd become used to the utility of it and had enough money to pay for it. I think BA made quite good money out of it, more so than Air France. BA's advantage is that a lot of those New Yorkers went to London as it is a very large financial hub, and less so to Paris.

To that customer base, the cost of not being able to hop across the Atlantic would have been pretty high. Things get really wierd in the world of large scale finance. For example, Tomas Bscher (apologies if any mis-spelling of that) bought a McLaren F1 back in the day and used it to commute from Cologne to Frankfurt. He was some hot-shot financier, and I remember reading an article relating how he considered that the price of the car had paid for itself in the time saving he gained on his commute (or at least, that's a good thing to tell the Misses!). It was effectively "free" motoring so far as he was concerned. He famously had a >200mph average speed in that car on his commute.

I fear that, if there's a large scale rise in supersonic airliners, the impact on ATC and airlane congestion could be severe. Concorde required "special" ATC treatment; it couldn't exactly afford to hang around in a stack wait to land, and would anyway have to stack somewhere else (because it burned a lot of fuel to fly slowly). If new supersonics require the same kind of ATC deal, airport capacity is actually going to drop if they're flown in even modest numbers. Whilst the problems of "noise" and "efficiency" might get addressed, what I don't think will change will be the physics of how fast such an airliner has to fly in and around the approaches to airports.

That's why some aircraft were swing-wing.

SpaceX resets 'Days Since Last Starship Explosion' counter to zero, again

Re: Oh Dear

Trouble with securing the pipe at more positions is making sure that the thing one has tied it to is itself not vibrating!

One possibility I've thought of is shock absorbers. Not the classic oil-filled ones for a car or a bike, but one using the same fluid as is in the tank.

For example, suppose the troublesome pipe is a methane pipe going through a lox tank, and the trouble sets in as the lox drains away leaving the methane pipe free to wobble. Have some shock absorbers filled with liquid oxygen between the pipe and the tank walls or some baffle structure. The shock absorbers would be fitted dry, and would fill up when the lox is loaded. The lox in the shock absorber wouldn't stay liquid for very long, but then it doesn't have to. The tank isn't going warm up whilst there's lox in it, and methane won't have to flow through the pipe when the lox is all gone. The shock absorbers might need some rudimentary lubricant or low friction surfaces (wasn't this what PTFE was developed for?), and could even be a wear-item to be replaced each launch. E.g. it could just depend on lox for some lubrication, and if it wears anywhere just make those parts from steel and have a magnet to catch the particles.

Oh Dear

Seems like it's still shaking itself to bits. It's progres of a sort - it got further - but this is going on for rather a large number of flights. Intentionally stressing a vehicle in test is one thing, but at the moment they've not bracketed a mission design / load / profile that doesn't lead to failure.

Most iterative approaches start off gentle to see it the thing at least works, then harden-up testing thereafter. It feels like they're trying to claim they're doing the opposite. But to me that smacks of over confidence.

I think what's going on is that they're slowly discovering that vibration modes all over the craft are severe, complex and difficult to solve. Solving those through iteration is going to take either a lot of material to dampen down the vibrations in pipes, structures, etc, (which will add weight), or an awful lot of analysis (which is not something Musk likes apparently), or a very long time and luck.

Worse, with vibration being mostly due to turbulent flow in pipes, success on one day is not guaranteed to lead to success on another day with the same design. The flow's turbulence is hard to predict, and not assured to be consistent (because it's also subject to vibration of pipes). A craft like this is going find all manner of exciting and thrilling ways in which it can tear itself apart, and it could take a looooong time before it stops surprising them with new ways of falling to pieces.

A lot of the success of such a program relies on progress, and enthusiasm. Musk has pretty much destroyed any credibility he had, and his team knows that (regardless of whether they're fans or not). If there's any hint at all now that money is getting tight, people will start to leave the project. With Tesla seemingly getting into severe problems, and Musk's performance over Twitter, there's likely a lot of wise money choosing to not get involved in any Muskian enterprises right now. Raising seriously large amounts of new money could be difficult.

China spawns an x86 supercomputing monster, with an AMD connection

Re: Interesting

Interesting that AVX-512 garners a reputation for being a power hog... I'm interested in how AMD's spin on it pans out; I can understand Intel's being a bit hot. AVX 512 is probably pretty good if one truly has a workload to stretch it.

SMT4+; we've been there before with the Inmos Transputers (well, sorta).

Microsoft revives DOS-era Edit in a modern shell

Re: "superseded Edlin"

Ah, not this one. The Teletype Model 33 had a travelling golf ball on a literal carriage - the paper stayed still. There was a lot of levers, solenoids, etc that shifted it, made it spin and tilt to the right letter, and then a big solenoid to make the golf ball smack the ribbon / paper very hard. I had the carriage return spring out of one of these for a while; it was a pretty strong spring!

Re: "superseded Edlin"

The whole street could hear the carriage return. Those were the days when a carriage return really meant something phsyically travelling at appalling speed and thudding to a stop when it got back to the LHS.

Re: "superseded Edlin"

Last time I used Edlin was when using a Teletype Model 33 ASR teleprinter on a comm port on PC XT clone (that didn't have a keyboard or display screen).

I can remember the experience distinctly. Strangely, I hanker after the teleprinter...

Apartment living to get worse in 5 years as 6 GHz Wi-Fi nears ‘exhaustion’

Re: Just....

It's not really the consumer's fault. The devices are supposed to conform to regs that say they should have a gap in their transmission spectrum to leave the amateur band alone. Thing is, the required gap is different in different territories. The manufacturers tend ship them set up for the USA, which doesn't work in Europe. The regulators don't act to enforce the regs, or aren't empowered to. Which is not the consumer's fault...

I recall reading that Ofcom were being sued in the UK for the lack of enforcement...

We shouldn't forget that Ethernet UTP is itself not perfect. It's only twisted pair, much like a phone line.

Fibre is the way to go.

70-knot winds so far blamed for yacht disaster that killed Brit tech tycoon Mike Lynch

Re: Acquitted...

There’s a bit of unresolved business here in the UK. The UK courts extradited him to the US on the basis of the case the US prosecutors wanted to bring. It passed some test of soundness here in the uk sufficient for the uk courts to determine that his liberty would be curtailed if he didn’t voluntarily go to the US.

So this raises questions about the judgement of the judges involved. They looked at the case and said “yep he’s got to answer for it”. Yet a US court basically throws it out as a load of old tripe. Not the greatest display of sharp reasoning by our courts.

The inference is that if anyone else gets called up by US prosecutors with a feeble case, the UK courts will pretty much roll over regardless. Which is the opposite of what they are meant to do.

With the death of Lynch there’s now no one who can make a claim against the uk system.

Re: Wrong Target....

Lynch was acquitted in the US, the whole hearing being over in very short order from what I remember. If speed of acquittal is any indication of how good a case was against him, one would have to conclude that the US court didn’t seem particularly interested in it…

Sudo-rs make me a sandwich, hold the buffer overflows

Re: Another day, another attempt to force this on us

Hmmm, I think you're over-optimistic about how many eyes are actually available to look at C/C++ code.

Re: Another day, another attempt to force this on us

You can have memory safety in any language you like.

The difference between Rust and C++ is that, with C++ someone else has to read your C++ very carefully indeed for them to know that it is completely memory safe, whereas in Rust it's simply a matter of looking to see if "unsafe" has or has not been used. Both can end up "correct", but it's simpler to demonstrate that state has been achieved with Rust.

"Suitability for purpose" is as much about the means of achieving an end result as the result itself.

Re: Another day, another attempt to force this on us

The supply chain for the first cars probably included horse drawn wagons.

Take a look at this video about the development of the original Shinkansen bullet train in Japan:

https://youtu.be/iYjFOYLAtoE?feature=shared&t=697

I've put it at a time point where you can see the first Series 0 power car body being shunted around by a steam train! Talk about big leaps forward. The whole video is pretty good, you get to see some of the hardware for the automatic train control system they built (in 1964!). Not much software involved.

Ironically, there's some Rust involved.

Re: Another day, another attempt to force this on us

Rust is not a serious competitor for C++. It will never compete in that space - ever. It will take margin away from Golang and Python then fall-back into single digits of interest.

I remember assembler programmers saying something similar about C... There's not many of them left.

Microsoft to preload Word minutes after boot

Re: More bloat in order to reduce bloat

Libre Office on Windows offers to set itself up in a similar way, when you install it.

I've literally just installed 7.6.2.1 on W11, and that's what it offered.

Re: Andy & Bill's law is alive and kicking

>So office apps on a Mac are by some distance the fattest, at 1 GB EACH for Word, PowerPoint etc.

Isn't that because all Mac apps come with all (supported) locales in one lump (a consequence of how Mac apps are put together?)?

Resource files for every single language under the sun which Word supports must amount to a large amount of data.

Re: IOW

When you install Libre Office on Windows, it gives an option to "load Libre Office at system start". I wonder if it's doing the same on Linux by default?

Linux in Excel? Sure, why not ruin both

Re: The idea of...

And interrupt vectors, and interrupt sources. All in all it could be a tad tricky to build the whole thing!

Windows isn't an OS, it's a bad habit that wants to become an addiction

From the article:

"All of this is reprehensible from an engineering viewpoint. Windows used to be terrible, then it got good, now it's getting terrible again because it is exempt from competitive forces."

Well that's certainly not true at all. Whatever competitive forces have or currently exist for Windows, that's been independent of whether or not Windows itself is any good. Macs have been a sensible competitive desktop / laptop option for quite a long time and has carved out a healthy slice of the market. Since then we've had Vista (bad), 7 (good), 8 (bad), 10 (good), 11 (well, I quite like it though I turn as much of the bloat off as possible)...

Under the hood, the Windows kernel has been pretty solid ever since version 7. For all its cosmetic faults, 11's kernel has brought in some solid security improvements. It also works just fine on ARM. Windows has been pretty good at maintaining software compatibility over long periods. Granted, they've had some howlers of patches screw ups, but MS arent' exactly unique in that regard...

Whether or not an OS is good or not depends largely on what one wants from an OS. The best one is always the one you yourself reaches for out of preference. Myself, I use all of Windows, Linux and MacOS interchangeably, and I also develop for a few more OSes beyond these three. They're all different and do things their own way. There's bits of all of them that I dislike, and others bits that I like.

AI-driven 20-ft robots coming for construction workers' jobs

Re: Solving the Wrong Problem?

I'm sure there are, I'm just pretty certain that using bricks is a bad idea in this day and age.

It's interesting looking at housing in Japan (which, I do). Modern buildings are either rock-solid concrete construction (mostly appartment blocks), or wood. The latter is considered disposable, and it's pretty standard when buying a house to scrap the old one and have a new one built in its place. That's the kind of approach one gets when there's a near certainty that a building is going to get hit by either an earthquake, tsunami, volcano, typhoon, land slide. Japanese housing isn't so great when it comes to things like thermal or sound insulation, but at least they don't hang on to it forever.

Re: Solving the Wrong Problem?

Sounds like an expensive way of not hiring a crane!

Solving the Wrong Problem?

Building a robot to build houses using construction methods such as timber, or brick piled up one by one is nonesense. These are ridiculous ways to make dwellings.

It's far better if the house is constructed as modules in a factory using modern methods and materials, and then transport these to the site and simply put in place. Huf Haus in Germany do this with engineered timber and other modern materials, and their houses are very nice indeed. Another is, building entire rooms out of concrete in a factory and then simply transporting them and stacking them up like Lego to build a house.

All sorts of ways better than piling up timber or bricks have been developed, but often fail in the market place largely out of fear of the new, of the "unknown". Rather than make a complicated robot that builds houses the old way, it'd be far better to build houses in a better way, probably with simply automation in a factory setting.

Devs sound alarm after Microsoft subtracts C/C++ extension from VS Code forks

Re: Sceptic

Because it’s open source and quite good?

In wake of Horizon scandal, forensics prof says digital evidence is a minefield

An Important Angle Overlooked

An important aspec that is widely overlooked is the role of the judiciary / courts in all this.

Down the decades, the courts have been repeatedly "hoodwinked" by suppliers of specialist evidence. There's many instances, e.g. Roy Meadow, the highly respect pediatrician who had a habbit of dishing up "evidence" to courts resulting in many women - recently bereaved after their children died of cot death - were jailed. In the earlier days of DNA evidence being used, the standard for "matches" was too weak, resulting in people being wrongfully convicted; that got corrected only when a wealthy convict was able to demonstrate that, whilst he was a close match to the evidence, he was not in fact an exact match (and therefore clearly not the person who had carried out the crime). There's been this post office scandal, where the courts have allowed an evidence *type* of a very dubious nature to be used *700* plus times. In Scotland there was the scandal about fingerprint evidence being made up. There's been cases in the Family court where social services armed with bogus "evidence" have repeatedly pursued families (there's one case in particular that hit the Sunday Times, and was resolved only when a French court granted the family asylum and protection from the British courts / social services). There's been many more.

Not once in any of these judicial failings has the courts intervened, and asked "hang on a mo, are you sure?". Leaving it to the prosecution and defence to "argue" (to the extent they're permitted) about technical or scientific is lunacy, and the courts do not even all the argument about such evidence to be fully explored (they know juries cannot follow it).

The most important thing that should come out of this is that the courts / judiciary themselves need to wise up on science and technology, and become more of an "inquiry" rather than an "argument". I think this is how things are done on the Continent, with courts being inquisitorial rather than adversarial. It seems to me that whilst our courts are adversarial, we're going to continue to have screw ups like this.

The fact that the courts have repeatedly been hoodwinked and that the courts / judiciary have not acted to prevent that happening again suggests to me that they're as culpable in the Horizon Scandal, and every other similar judicial scandal. If they continue to refuse to acknowledge that they have a problem, they are not doing their job.

Location Data…

…has got to be questionable in this day and age when GPS spoofers are prevalent. Location logs can’t tell the difference.

Dot com era crash on the cards for AI datacenter spending? It's a 'risk'

It’s a certainty…

Satellite phone tech coming to your mobe this year – but who pays for it?

Re: This is What'll Happen

It's effectively a law of comms that - if there's enough people with enough money to be worth supplying communications to in any area, there'll already be a 4G/5G base station there. What's left over has a very low market dollars per square mile on which to found a satellite network. There are distortions to that of course, and the US is a good example of where odd market conditions have meant that there's precious little competition (and thus massive underinvestment) in terrestrial comm for areas where there's a reasonably high concentration of money on the ground.

The pricing is indeed going to be interesting. Though depending on exactly what the technical solution ends up being, the two otherwise "identical" plans probably aren't in some detail that truly matters. Suppose it becomes possible to supply high-rate data connections from satellites to mobile phones, and the same in reverse. The temptation for the vendor would then be to provide that in preference to cellular coverage. Rather than maintain those costly base stations, just dispose of them as and when, the satellite coverage will fill the gap just nicely. However, transmissions to a satellite from a phone will always require more engergy than the same amount of data sent to a cell tower just down the road. Your battery would go flat faster. It doesn't matter if your primarily downloading; those IP packets won't ACK themselves...

I think that - as a low data rate, SMS-style service - it makes a lot of sense. Emergency messages can be life savers. But, as part of an emergency service, that's not really something for competitive, profit-motivated market dynamics.

This is What'll Happen

My 2p...

Those networks that roll it out "as standard", part of their packages, what they'll find is that it largely goes unused. There will be sporadic, occassional use, or very concentrated use (such as in those Pacific islands mentioned in the article, where I'm guessing cell, WiFi and broadband coverage is a bit thinner). Then they'll wonder why they're paying for it if no one is using it, and stop providing it.

Those networks that have it as a paid add-on are likely to find an initial flurry, and then they too notice that punters aren't buying it. But, at least they're not losing money on the deal.

The problem with all this kind of thing for retail mobile telephony is that, having spent so much time building out huge networks of cellular coverage, and with most punters also having access to broadband or WiFi at home, work, in some cities, on trains, buses, and planes, there's comparatively little call for retail satellite mobile services. And if the tech allows them to start offering fat broadband pipes in preference to cellular coverage, it'll cost quite a lot of money and will probably flatten phone batteries pretty rapidly, and won't work indoors anyway (which is where a lot of people in air-conditioned countries spend their time anyway, along with a WiFi hotspot). No one in their daily lives wants to have to sign up to each and every WiFi service in whatever buildings they're walking into, simply to get coverage. Cellular service coverage inside buildings is a major aspect of its service provision, and also the reason why the original networks are focused on the 900MHz band which is the ideal compromise for mobile antenna size and good penetration of structures.

Things may change, of course, but I strongly doubt it.

RISC OS Open plots great escape from 32-bit purgatory

It's getting harder and harder to not use 64bit ARM / Linux, or other similar ecosystems.

For example, NXP now do a chip, an ARM SOC, that (at most) draws 1Watt, measures 8mm x 8mm, and runs a full 64bit Linux, with PCIe, Ethernet, USB, all manner of buses on board. When one considers that it'll power-state switch itself down far below that 1Watt when nothing's happening, it's getting pretty hard to find reasons not to use them. Sure, one can stick to microcontrollers for really cheap / basic stuff, but as soon as one wants to connect it, you're far better off with the ARM SOC / Linux combo.

Plus, it's far easier to find Linux developers (who can include Java, other high level languages) than it is microcontroller programmers. And it's far more productive.

Musk's xAI swallows Musk's X in ego-friendly, all-stock deal

So Where's the Debt Gone?

Musk borrowed a ton of cash to buy Twitter, and I presume piled that debt on to the company. Is that's now the responsibility of xAI and its investors?

Now Windows Longhorn is long gone, witness reflects on Microsoft's OS belly-flop

Re: WinFS = Document Management System not invented here…

What you're proposing would - if WinFS were to do what it is described as doing - mean that a "document" would end up as innumerable individual files on an NTFS file system, each one containing just one part (e.g. one span of text with the same formatting) of a document, and WinFS having an index of files that comprise "the whole document". If one were to offer a parallel interface to files - the normal one and the WinFS one - the normal one wouldn't show a "file" that is the document at all, just the innumerable individual files, one for each object in the document. I don't think that would have made any sense.

WinFS did in fact store some of its data as files on disks - but that was limited to unstructured data, and the unstructured part of of semi-structured data (such as images - unstructured bit stream plus structured metadata). But for structured data - they way it was intended to be used - that was all stored in tables in the database. What MS seems to have been aiming at was that all apps would expose the structure of their data to WinFS, and interact with it only through the medium of structured data storage interface on the database.

We've never really seen in general release anything that actually achieved this way of storing data. Gnome had a similar project for a while (GNOME Storage), but that seems defunct. Had WinFS or GNOME Storage been achieved, and all apps dealt with it purely through the medium of structured data, I suspect it could have been pretty performant. RDBMSs can be pretty performant when used well, and an awful lot of functionality (such as multi-user editing) could have become the default for all applications without the application developers having to implement it.

Bill Gates is said to consider that the failure to complete and deliver WinFS was a matter of regret. And I can understand why. The motivation - bringing universal meaning and standardised structure to all your data - was sound enough. Instead, what we generally have on all platforms (Windows, Mac, Linux, iOS, Android, etc) nowadays is myriad search bots, databases, files, other files that are also databases, all with their own code base, all tying up a lot of CPU time, all with their own vulnerabilities. On Windows alone I can count several - the Drive C: NTFS, the Windows Registry, the numerous databases that make up Active Directory, the file search / indexer's own database, the database storing local user account information. Heavens knows how many there are on a Mac. On your average Linux there's quite a few too; /etc/passwd, /etc/group, the ext filesystem, systemD's journal, /etc/shadow, /etc/hosts, etc.

Replacing the lot with just one database engine for the whole system (which is kinda what WinFS was heading for) makes quite a lot of sense. It'd probably be faster, and there's less room for security vulnerabilities across the system.

Re: WinFS = Document Management System not invented here…

That's kinda missing out on what the whole point of WinFS was all about. Sure, in NTFS you can tag files with metadata. But that still "files". With WinFS, there weren't even files. There were merely objects, a collection of which could make up what would conventionally be called a "file".

As I've said in a comment below, one can look at something like a Word document as a collection of objects of various types, ordered in some way. All that the application is doing is rendering objects to screen in that order, and allowing the user to create / edit / delete objects and move them around in relation to one another. In WinFS, the objects would each be stored as a record in a table of the right sort, and a search pattern defines the "document".

One can look at something like a text file as a database of lines of text, with the proper order of the records being the order of the records in the database. An RDBMS could store a text file that way, a list of record indices could define the "file", and the whole thing could be given a POSIX open / read / write / close API interface to make the whole thing look like a file on disk whilst in fact it is not. In fact, looked at this way, having an "insert" function for a text file would then be efficient, because this would be just a matter of adding more records and updating the list of record indices. Whereas today we're still stuck with re-writing the whole file from the insertion point onwards.

It wouldn't have made any sense for there to be a user-accessible NTFS underpinning WinFS. A lot of the benefits would have been realised only if every single application used it "properly". It would have been necessary to expose a file system API (open, read, write, seek, close) upwards for those apps not year prepared to talk to an RDBMS, and I suspect that at the time the performance could have been a bit rubbish.

I find the idea appealing in the sense that in that it gets right to the heart of what storage is actually for. Though possibly, I'd have not found the performance appealing!

The idea has come back to life, I think. I'm pretty sure that Web apps such as Google Docs are dealing with a database of records, rather than with a single file containing the whole document bytestreem.

Re: WinFS = Document Management System not invented here…

I'm not sure why you think that. A decent database engine has a built in permissions system, possibly down to the level of individual records, in the same way a file system has a permissions system down to the level of individual files. There's not much difference, really.

A lot of the code, utilities, practises and ideas are very similar.