Re: Yes, its you.
Um, yah. Rob Pike and Ken Thompson are wee ankle-biters.
Tony Hoare is 20 years older, and is entitled to call a 52 year old Rob Pike a youngster!
1926 posts • joined 23 Apr 2008
Um, yah. Rob Pike and Ken Thompson are wee ankle-biters.
Tony Hoare is 20 years older, and is entitled to call a 52 year old Rob Pike a youngster!
Is using pthreads REALLY so hard? There seems to be a lot of noise about the latest flavour of the month concurrent languages but in reality all they do is prettify (and arguably simplify) threading syntax and control then make the same underlying calls to the OS threading system. They don't actually give you any more power.
PThreads are harder to get right (from the point of view of being sure there's no potential for deadlock, lock failures, etc). Shared memory, semaphores, etc can be fearsomely difficult to debug, etc.
CSP in particular makes it easier to get it right, or at least if you don't you're guaranteed to find out as soon as you run your system. If you've written it such that it can deadlock, it will deadlock.
Also you can do some process calculi maths and prove the correctness of a CSP design theoretically. That's not something you can do with pthreads, shared memory and semaphores in anything but the most trivial of cases.
As for more power, one has to be a bit careful. Pthreads, shared memory, semaphores all assume that it's running on top of an SMP computer architecture. Whereas CSP is quite content to exist on NUMA architectures. What Intel actually give us is, effectively, a NUMA machine with SMP emulated on top. Thus software that's more NUMAry in operation can be kinder to the Quickpath interconnect between CPUS and the shared L3 caches.
And because CSP is NUMA friendly, it's quite straight forward to scatter CSP proecesses around a network and scale up (though Go and Rust don't AFAIK do this for you). That's a complete no-no with shared memory, semaphores and pthreads.
If you like Go's concurrency you may also like Rust. That too gives you Channels which can be rendezvous mechanisms.
Rust is beginning to look good because it is suitable as a system's language. There's a whole lot of youngsters at work getting excited about it.
To us old timers the re-emergence of CSP channels is a delight! It's definitely the easiest way to do parallel execution. Rust'll save me having to implement CSP in C++ for myself every time...
Parliament produces an annual report on surveillance activities, and the court case results and transcripts are openly available. Go read!
But yes, Trump's Muslim registry is an abhorrent idea.
I suspect you're too young, of the wrong nationality, and lacking in scholarly study of history to really know what horrible bastards the STASI really were. Count yourself very lucky indeed to not be living somewhere anything like the former GDR.
From the article:
"Her position is ironic considering her first act in government, way back in 2010, to repeal Labour's Identity Cards Act 2006. May claimed the Act would not “keep us safe without intruding on civil liberties” and continuing that while “some data storage is essential,” in her explanation to Parliament, “these events do not point in the direction of a massive expansion of the surveillance state, which ID cards would necessarily involve.”
The ID card scheme's planned use of biometrics was mathematically flawed. It was never going to be able to deliver what was wanted because the biometric ID part of that was going to be insufficiently reliable. One might therefore argue that binning the scheme was actually a well measured move on May's part, as nothing good was going for come from having all that data on file, all that equipment in place, etc.
"David Davis, currently Brexit secretary, so disliked the PM's last attempt at a Snoopers' Charter that he with others took the country to the EU's highest court. The resulting judgment seems to put much of the UK's new data retention regime to challenge."
Ah yes, David Davis, the man who resigned from the then shadow cabinet in protest at something that the then Labour government was doing. Can't remember what it was all about, but the lasting impression his protest made was "idiot": it's a Shadow Cabinet politician's job to criticise government where necessary, something you cannot do if you quit. A hollow gesture at best, and at the very least it was poor judgment.
Anyway, the true irony is that whilst many may look upon Europe as a means to moderate whatever 'excesses' one might perceive to be built into the IPA, some (e.g. France) and probably many other European governments are planning or implementing similar laws.
The wave of attacks on various blameless and/or beneficient European countries (what's Belgium ever done to annoy anyone? Chocolate?) has finally woken them up. We're all at risk from such attacks, and a policy of non-involvement in the world's trouble spots and a laissez-faire attitude to what's happening in communities in your own capital city is not good armour against them.
The EU and ECJ may witter on about open borders and human rights, but European politicians now realise that there's elections to be lost through inaction and failure to prevent repeat attacks. Everyone remembers that the ruling party in Spain lost the next general election immediately after the Madrid train bombings. There's nothing like the prospect of losing their job to make politicians engage in hard realpolitiks.
Anyway, my point about Britons seeking redress in the ECJ about what they perceive to be the excess of the IPA is that if any government tells the ECJ to back off it might not be the UK government that ends up doing that.
What would be truly excessive is not having laws (or having very weak laws) about what the state can and cannot do in the area of domestic security, and then making it up on the spot or doing things with no legal top cover in response to the popular backlash against a government hitherto inattentive to such matters. For example, France is still officially in a state of emergency. Who knows what they're doing under that cover. And last time I was there (Nov) there's still troops on the streets of Brussels. This is far from "normal".
An interesting listen on such topics is the Reith lectures given by Eliza Manningham-Buller, former head of MI5.
All things considered I'd far rather have open laws on domestic surveillance and security than opaque states of emergency, or no laws at all. Whether or not they're effective laws or not time will tell.
No matches being applied to the pyre from this here IP address.
The only thing that I'd disagree with Andrew on is that 2017 would be the year that it all started to unravel. I'd say that it's already started with Google effectively abandoning their self driving car effort, all the way back when in 2016. Self driving is probably the most ambitious AI challenge that was being contemplated and, unsurprisingly, it turns out they couldn't make it work and don't know how to go about improving it. Apple have also wound their necks in on their equivalent ambitions. Uber look positively behind the times in persisting with their endeavour, which has only just got going. And with the big tech-companies running into difficulty, the threat of them disrupting the traditional auto manufacturers' business is diminshed, so they too will drop the topic with a quiet, unpublicised thud.
Unfortunately, juding from various electoral results across the world, one might also have to conclude that human intelligence also started to unravel quite badly in 2016 too. Rise of the machines? Hardly. What we actually got was the Rise of the Trump.
No. Not even MS could be that mad. Ooops, perhaps that'd have been better unsaid, lest they get an idea...
All those wasted, fruitless and ruinous years of shoving the PC industry down the tubes with comical, immature user interfaces that no one on the entire planet wanted. They'd already got what had to be one of the best desktop UIs (Win7) ever. If it's coming back, it feels like it's doing so through shear osmotic pressure rather than any kind of coherent plan. Whatever next - Aero? Carousel?
If they did something now that had Window 7's UI, none of that advertising / snooping nonsense that is the final ruin of Windows 10, and a tabbed Windows Explorer, I'd happily pay £100 for it.
It sounds like the better thing to campaign for is freedom of servicing (such as exists by law for ordinary cars). It's a law change that's needed to make that happen, not access to software source code.
It sounds like Deere are fully exploiting the lock-in opportunity offered by the law as it stands, which is kinda dangerous for them. First, it could / should provoke a change in the law (which might include some punitive retrospective measures against manufacturers who had been taking the piss). Second, it leaves the market wide open to a more enlightened manufacturer to come in and tempt the customers with easy servicing options.
Because anything manufactured by industry is certain to be built, tested and operating according to all the rules? And if the implementation is kept secret, how exactly could you be sure of that?
Er, by black box testing it. All designs have to pass a bunch of tests, and unless they pass they cannot be sold. Who cares what the software is like if it passes the required tests?
Given the vast number of cars that are sold world wide one has to conclude that the regulatory oversight of the industry is highly effective; there's comparatively few defects that escape notice, and the regulatory recall system works well to fix those that crop up later on.
Didn't work for emissions. Why should it work for anything else?
VW emissions cheating was found by black box testing, not examination of the software. It was an imaginative step by the researchers who found it, and it's resulted in the regulatory inspections being tightened. So there's now even less reason to examine the source code than before. In the same piece of research they found that a number of other vehicles did perform properly, again established without access to the software.
The UK is comparatively unusual in that individuals are allowed to design and build their own car from scratch and drive it on the road. The *only* thing you have to do is get it past a vehicle inspection. It's quite a big inspection, but it's affordable (costs a few hundred pounds). Some people have even done this to the extent of designing their own engine. There's no requirement to show that all your own software has been written according to MISRA rules, etc. In, say, France it's different; an individual has to go through the same crash test regime as a major manufacturer - prohibitively expensive.
So in short, the source code is irrelevant. Even if you had it, that makes no difference to get tests a vehicle has to pass.
The only thing that the source code might usefully do is allow servicing monopolies to be broken. But getting the law changed is likely to be a quicker and more effective way of achieving that.
They're basically saying that they're not going to expose themselves to potential liability by releasing their source code. Can't blame them for that.
No matter what wording they might put in release notes, releasing the source code at all could be construed by a clever lawyer (especially in the US) as being equivalent to "do what you like, you can't go wrong". However in these days of throttle-by-wire, brake-by-wire, etc that's far from being true. They really would be running the risk of being sued by the victim of someone who'd modified it.
Getting the law changed on open servicing for farm vehicles, plant, etc. sounds like a far better bet. Putting campaigning effort into that would be more productive that cocking about with copyright laws.
...Whether anyone at Apple ever bothered to turn one on more than once in development...
As I read these comments a spam email from my provider has arrived plugging the benefits of a smart meter!
Been ignoring them for months now. Some of my neighbours have had them installed, seems like the fitters have taken the opportunity to suggest lots of expensive and entirely unnecessary remedial work on the houses' electrics. 1990 houses. Not ancient at all, no rubber insulated wires here, etc. Con.
But only until you observe it, and then it's pot luck as to whether you can read it or not!
Also missing is by comparison from El Reg's analysis of military drone incidents, is why it was decided to be solely flown on autopilot. Where's the oversight and the person with the "abort landing" button in this?
USAF Predators are (were?) hand-landed, US Army's we are auto landed. Something to do with letting USAF personnel think they had a 'flying' role. Guess what, USAF lost loads in landing accidents...
There's a big difference between things like the sturdier Predator and this flimsy thing from Facebook. Predators are designed with good endurance, but Aquila is all about staying up for a very long time indeed. But military UAVs, despite being sturdier, aren't that fond of nasty conditions either. Strong winds, poor visibility will ground them.
A human pilot on board (e.g. On something like an A320 / 737) is able to land in far worse conditions, and are able to deal with lots of faults, and are pretty good at improvising landings on rivers, etc.
@Orv, yes that's all sensible stuff. It's similar in the aero engine business too - GE, RR, PW are deadly serious competitors but assist each other to help out on safety issues. It's a mature industry that operates well for the benefit of passengers and suppliers alike.
I think that California is also being very sensible by permitting and regulating tests and publishing test results. It's forcing the nascent self driving car industry to behave with maturity, and most importantly of all it's not letting PR control whether or not these things go on general sale.
That's definitely caught out Google's team, whose enthusiasm for self driving has taken a dive following the publication of their test results. One wonders what would have happened had publication not been an obligatory part of the process.
If Uber start chipping away at that edifice of rationality then, as you say, that could ruin the current beneficial status quo.
My own point of view is that we'll never be able to develop a self driving car that is demonstrably as reliable as the best human driver (no point being as good as the average driver, most of us would be worse off). We simply don't have the means to even write down in detail what it is that a human driver actually does and can do if required. And if we can't even write that down on paper, how can we ever test a self driving car and show that it is equal to the best human driver?
There will always be some unexpected circumstances that defeat a self driving car (anyone tried one on a dark foggy night on narrow twisty British lanes with black ice here and there, lots of leaves falling from the trees? Thought not...), and you won't want the car getting into a panic when it's got only your kids on board.
At best we might get an elaborate cruise control thing, but then there are human factors concerns about drivers who spend very little time actually driving suddenly being expected to take over in circumstances that by definition will be challenging. Just look at the problems surrounding Tesla's Autopilot... I'm hoping that State regulators will think very carefully about that aspect.
Trouble is any kind of reasonable unlocking scheme isn't going to be very compatible with taking snapshots in those fleeting moments of inspired opportunity.
"The answer to bad speech is not censorship, it's more speech calling out fake news," he said. "We need to spread the idea that critical thinking is important."
Er, am I alone in thinking that that's an ambitious aspiration?
Also I'm not entirely sure what fake news is, or rather more specifically, what real news is. People wring their hands about the traditional press and its 'real news' being eroded to irrelevance by online fake junk, etc, but then again the traditional press includes the Daily Mail, the Murdoch-owned newspapers, the Telegraph, Express, Guardian, etc. We 'know' that a lot of their output is designed to suit their owners' views and is far from being 'true'.
Even broadcast news is questionable - in the UK it's pretty much the same stories in the same order with the same-ish spin no matter which channel one tunes into. TV journalists are told by their editors what sort of story to get, and they choose what interviews with "members of the public" are actually used. And even if one thinks that the news channel is playing it straight, one often has to wonder why so many placards are written in English and displayed in so many protests in so many countries where English is not widely spoken.
I've long ago concluded that the only reason something is broadcast or printed or Web published is because someone somewhere wants you to absorb their point of view to further their own agenda. Editors mostly, press officers too. The difference being these days that almost anyone can be an editor online, and social media 'trending' popular content brings it to the fore. Fake news? 'Real' news outlets are just as capable of deliberately misconstruing or misrepresenting or imagining 'facts' to suit the agenda of the owner of the newspaper, TV channel, etc. For example, look at the BBC's abysmal role in spreading the demonstrably false scare stories about the MMR jab. Balance? Balance my arse - sensational stories bring high viewing figures, broadcast and revel in the numbers, that's their only motivation. On this particular topic, Brian Cox pointed out that the BBC's 'balance' weighed the entirety of scientific peer reviewed wisdom against the opinion of a man not accepted as an authority in the field with no peer reviewed papers to back him up, and judged them to be equal. Oh dear.
Anyway, so almost no one involved in the creation, dissemination of 'real' news is truly an honest broker. So why is 'fake' news significantly worse?
About the only exception is Private Eye, which is pretty good at putting out mostly facts (rare mistakes) and mostly letting them speak for themselves. And the tech press like El Reg does a good job of sticking to tech news.
...when and on what planet did anyone think that information was going to treated well and respected?
Well all across Europe there's quite strong data protection laws. Breaches like the ones reported in the US would result in some stringent fines. In the UK this can be up to £5000 fine per data item lost / breached / exposed. Consequently Megabreaches are potentially veeeery expensive (millions x £5000), and that rather focuses senior management attention on data protection within a business. Basically you don't collect data unless there's a genuine business need, and you don't make it available willy-nilly within a company, etc.
But Uber is USAian, so like many other companies they're rather uncaring able their customers' data, what they collect, etc.
I had no idea they were designing new space suits. I'm somewhat mystified as to why existing designs aren't acceptable. Do you suppose it's simply a matter of IPR ownership?
Oops. For 'billion', read 'million'. Hadn't had an early morning cuppa.
Anyway, to return to the point, for any half decent payload (big TV satellite, big comms sat) the owner would have paid anything up to $1billion building it. ENVISAT cost a reputed $2.9billion. Uncle Sam will quite happily spend $1billion on a spy sat. The one SpaceX blew up a few weeks back was a comparative tiddler money-wise, only $200million. There's a good book here, page 99.
In comparison, even an "expensive" launcher like Ariane will be ~$100million ($150million if one's satellite takes up the whole launch payload), Wikipedia's page is quite informative on the matter. And SpaceX are seeking to chop costs like that into tinier pieces.
It's been a long time since launches were more expensive than building big satellites.
Manpower. SpaceX seem to have no ambition to reuse a booster more than 1 or 2 times. That means that for the foreseeable future they will have to retain the skill base, currency, and capacity for something close to full rate production (assuming that a few don't make it to landing as planned). These are not talentless people who can be picked up off the streets at short notice, these are hard-to-find guys and girls who once you've got them you have to retain them, pay them, keep them busy, etc. It's not like they're necessarily jack of all trades either - they're specialist welders, machinists, etc, all highly specialised in their individual fields and not easily transferable to different roles. A lot of that skill base may very well be vested in suppliers' workforces, but it's there same problem all over again and the costs of dealing with it will be passed on to SpaceX and anyone else buying from the same supplier. Same for the factory - whilst there's a single final assembly building, there's a myriad of smaller plants all over the place that have to be kept operational if they're to build just one booster per year.
And on top of that SpaceX would need to have the refurb staff too to recycle the one's that did make it to landing. If, and only if, SpaceX can re-use their first stage many many times (e.g. 50 times) is it worth building a fleet and then standing down the production line.
This problem aflicts every major large engineering production project. Fighter jets - the unit cost goes up as the government's order shrinks for exactly the same reason. To get an empty cardboard box from Lockheed or SpaceX would cost almost as much as putting an F22 or Falcon 9 inside it. The F16 production line is about to close, and once gone it'll be veeery expensive to bring it back should anyone want a new build F16. The overhead of having the means to produce things this complicated yet not making them is almost as expensive as making them regardless.
And since you mention cars, it costs Ford / GM / Merc / etc. around about $1-6billion to develop a new car. Which is why there's so much platform sharing going on these days. Once they have developed it and set up the line they can churn out millions of cars for a couple of thousand dollars each (if that), but that first one is $1-6billion. Take a look at this.
Anyway, I'm only reinforcing SpaceX's own pronouncements on the matter, covered previously here on El Reg and elsewhere in the press. Take a look at this: SpaceX were talking about a 30% discount, tops. Even that seems ambitious to me.
So lets see... Say SpaceX charge $50million for a fresh launcher, 30% discount for a second hander = $15million saving. $15million/$200million = 7.5% of the price of the satellite that got blown up. That's pretty small beer. In the grand scheme of things, if one could ensure the success of an investment in a $200million satellite by spending an extra $15million on a fresh launcher as opposed to "taking a chance" (no-one really knows yet what reliability SpaceX can achieved for re-used first stages), one (or more likely one's insurer) probably would spend that extra. If it's Uncle Sam who's just spent $1billion make a new military satellite, $15million is really small beer indeed.
They will get to re-use one of their first stages one of these days, but the money saving isn't as attractive as all that to launch customers (or their insurers). If SpaceX can show that it works "as good as new", which I'm sure they'll manage to do one way or other, then it would become a no-brainer. But just at the moment it'd be a brave customer to bet their own enterprise on a small saving.
I'm guessing that they're having trouble convincing a customer to go with it. For various reasons re-using one isn't as cheap as all that, and so the financial incentives for a customer aren't that great compared to taking a brand new one. If they've spent a few $100billion on their satellite, risking it all for the sake of a few million extra for a brand new launcher is almost a no-brainer.
SpaceX may have to launch one or two on test flights to show that it can be done.
The reason re-use isn't as cheap as all that is because SpaceX still have to pay the fixed costs (manpower, supply and support contracts, facilities, heating, cooling, lighting, land, etc) of maintaining the ability to manufacturer new ones. That fixed cost is far more than the price of the materials that get thrown away every time they destroy one. Only if they become massively re-usable can SpaceX economise on those fixed costs.
That's probably what is going on, but that should be the kind of thing the devs assess in debug back in the shop, not live on everyone's phone where it's purpose may be misconstrued (as I did at the top!).
After all, discovering that one's app's power consumption is terrible from live telemetry back from users is bad for business and makes it look like one couldn't be bothered to check it oneself because shipping.
Of course they're going to charge more money to someone late at night whose mobile is close to switching itself off!
In fact, the more pointless computation the Uber app does the more battery it uses up and the more likely the "customer" is to be desparate to get home.
Sounds very similar to VW's emissions controls software!
I know a pub in deepest darkest Devon that had no cash register, just an ancient wooden box. The VAT man was unhappy with this, certain that they weren't paying enough VAT.
So in came a modern electronic till, and a proper sales record presented to the VAT man the following year. Turned out that they'd been massively over paying their VAT for years, and got a healthy rebate. Drinks all round!
Ask yourself this: how would the average Chinese citizen react to hearing their government actually benefited a foreign company over one of their own? Outrage! (and as calculated by the party).
It's more complex than that. Huawei, a Chinese company that does actually come up with good ideas of its own and makes good money selling them all over the world, has problems with their products being faked and flogged by other Chinese companies. There's not much patriotism in play, it's all about making as much money as quickly as possible no matter what the risks or consequences. Hence the various food contamination scandals, etc.
The communist government is quite capable of controlling what people think - it controls the media. A few years ago they stirred up some anti-Japanese sentiment for some crazy reason or other, riots, Japanese cars being burnt in the street, etc. Whatever it was all really about escapes me (there was some WW2 angle used as an excuse) but it went too far; the Japanese auto manufacturers started closing their factories in China, put a lot of people out of work. Funnily enough the riots stopped just as quickly as they'd started but it was too late. A lot of Japanese manufacturers decided that operating in China was more hassle than it was worth.
What makes this all very frustrating is that there's already perfectly good solutions to the leap second problem.
If OS developers wrote their OSes to use International Atomic Time instead of UTC as their base timescale, the OS would never need to deal with a leap second.
And there's perfectly good libraries for converting TAI to, e.g. UTC that already handle leap seconds, can do accurate time calculations, etc. One such example is the SOFA library from the IAU.
Like everything else it cannot predict leap seconds, but an OS is already well placed to receive library updates as part of its regular maintenance. Why not this one too? And if every developer used TAI instead of UTC to represent time values then all their calculations would always be correct, with conversion to UTC for display being the only thing that'd be wrong in the absence of updates.
The problem as always with something like this is clinical responsibility. If you make something that can tell whether or not you have some condition or other, it is saying either "yes, you have it", or "no you don't".
The problem faced is that if your box is even slightly subjective (such as a neural network output), you cannot really afford to have it say anything substantive. So the answers it has to give have to be toned down to "maybe, go see a doctor", or "maybe not, go see a doctor". In which case, what's the bloody point of not just going to see a real, qualified human doctor in the first place?
Not toning the answer down and giving a straight yes/no answer means you're accepting clinical responsibility for the accuracy of the answer, and the resultant liability for those occassions where your box's answer turns out to have been wrong. A false positive upsets patients, and may lead to damaging and inappropriate medical intervention. A false negative may kill them. If you've made claims of complete reliability, you take the blame for that.
So whilst their system might have a strong performance from a statistical point of view, it doesn't amount to anything practically useful at all unless Google are actually willing to accept the liability for the system's performance. I can't see them doing that.
The same's true for a human doctor, but they can get insurance cover.
Perhaps not all of this code is safety critical. Perhaps most of it could be banged out over a weekend, in the usual manner.
That's basically what the self driving car types are doing. Their approach is to get enough cars running incident free for long enough that they can get approval through mere statistical argument.
It's deeply worrying that this approach may actually be accepted by regulators. Unreviewed, unprovable safety critical code with no triplicate redundancy? No thanks. A statistics-based approval ignores the possibility of a systemic date-sensitive bug lurking unnoticed in the code base that will do something very damaging at some point in the future.
Fortunately it seems that their best efforts so far are a long, long way from being statistically acceptable.
In fact, the Japanese built an entire airport on a man-made island that is sinking at many centimeters per year... and all of the structures there are built on jacks. It's actually quite remarkable. So if something sinks more than expected, that doesn't mean the safety margin has been exceeded.
Kansai Airport has sunk far more than expected, and there is some nervousness that it won't stop before it sinks beneath the waves. The building jacks keep them level, but won't keep them dry.
The loss of Kansai airport would be hugely problematic; they're planning on closing Itami airport in nearby Osaka, leaving just Kobe's small airport as backup. Basically the whole region needs Kansai airport to stop sinking.
Data#3 look pretty dumb here. No employer anywhere is entitled to know everything about one's personal life and history. Sacking them subsequently when they've already asked whether the company needs to know of specific aspects about their personal life is pretty crazy.
They should ask themselves, would they sack someone if they found out they had liked the band Genesis? Many today would consider such a past to be fairly appalling, but it's not a sacking offence. Obviously if they're still a fan, well that's different...
I know, but the first flight of the first A350 was just as uneventful. And the A330. And pretty much everything else that has been flown in recent decades.
Back in the old days it was definitely more exciting. Bill Parks, Lockheed's legendary test pilot, demanded and got extra (extra! Ordinary danger money was already part of the package) danger money for taking Have Blue (the F117 prototype) up for the first time it was such a ramshackle assemblage of second hand junk. Hugely successful though.
There rarely seem to be any challenges these days.
In the old days the test pilots would strap themselves in, light the fires and take the aircraft up for a quick circuit and get it back down ASAP before one of the many items that are clearly wrong / broken / rattling / wobbling / leaking / overheating bring it down in a smoking ruin of bent, shattered and charred aluminium. A lifetime of excitement compressed into 5 minutes of sheer exhilaration, possibly ending in some parachute time.
Nowadays it's normal to be able to take the plane up for hours on the first flight and fully explore the flight envelope. One wonders what the remainder of the flight test campaign is really for these days.
Of course, this is a good thing.
Anyway, congrats to Airbus and the wider aviation community for getting it right so often. It's become so normal that we moan about delays.
They kinda did sort their shit out back in 2008/2009. They showed off win 7 + office recompiled for ARM at a trade show, and everyone though "MS is really getting to grips with ARM". And then instead they did Win RT, Windows 8, and it all turned into a debacle...
Windows has traditionally been reasonably re-targetable, they just failed to spot where things might go with ARM. They could right now do it, but with UWP targeting the non existent mobile platform (Win 10 mobile) they're still missing the point.
Apple are dropping their WiFi product line, Google are busily trumpeting their own that's not on sale quite yet?
One's gotta wonder what Google see in it that Apple don't. Slurping opportunities?
America's heading towards recession. That'll do no one any good.
Calling for regulated security on IoT devices is, well, likely to have consequences more far reaching than anticipated.
For a start, when is a CPU + memory + NIC + software an IoT device, and when is it just a computer or smartphone? They're all potentially involved in home automation, especially if you consider the app as being part of the IoT system.
To illustrate the difficulties of trying to make a legal differentiation between IoT and non-IoT, consider the Raspberry Pi. IoT device? Yes. Computer? Yes. Router? Yes. Server? Yes.
So you cannot reasonably apply a bunch of regulations to an IoT device that then don't also apply to smartphones, computers, home routers, smart TVs, back end services, the entire Internet, Thus if the law required IoT devices to meet minimum security requirements, receive regular updates, etc they'd have to apply to everything else too, otherwise there'd be no point.
That would be a problem for Android in particular.
Perhaps they're getting nervous of the direction in which Google are taking Android. The version on the new Pixels seem to be hinting at Google heading in a Google-only direction. If that's true Samsung will need a plan B.
Tizen doesn't seem to me to be a very good plan B.
It's quite interesting to see what the Chinese have done with Android. They've successfully taken it forward and, with a degree of government backing, supplanted Google's own services with their own domestic providers (Baidu). It's a flourishing ecosystem, and not being involved in that makes it very difficult now for both Google and Apple to get market traction there. China succeeded because it's big enough that their own local monopoly could do its own thing and ignore Google.
Doing another "China" would be a better plan B. Could Samsung do the same thing, provide their own replacement services? Probably not. Google's walled garden already encompasses most of the Western World. There needs to be government enforced separation of OS and services and open competition between providers (aka Break Up Google!) before another mobile OS can survive. BlackBerry have tried hard (excellent services, excellent OS, market failure), and it didn't stop the juggernaut. Microsoft too. Even Apple are losing out. BlackBerry may survive because, in their recent adoption of Android, they've decided to climb on board rather than be squashed by the tires again.
Musk made (to my astonishment) an interesting point. He noted that if all manufacturing were automated, we'd have to pay a basic wage to everyone just for living. That'd be especially true if all agriculture were automated too.
That'd be either extremely unhealthy, or very far sighted, or both.
A trade war is no different to a military war; you can't win, you just hope to lose less badly than the other guys.
The move of manufacturing from the USA to China stems from the fact that company chairmen and directors have a fiduciary obligation to maximise shareholder value. This is not a trivial thing, it's deadly serious, especially in the USA. In the USA you can end up in deep personal trouble (e.g. jail, ultimately) if, as a director, you fail to improve value through gross negligence.
One big way to improve value is to make manufacturing cheaper. China was cheaper than the USA.
Now if, 20 years ago, a company chairman, CEO or director had refused to relocate manufacturing to China they'd have been laughed at, probably sacked, or (depending on how they'd 'lied' about the reasons not to) sued / prosecuted / jailed. Saying back then that concerns about long term geopolitical consequences were a reason not to off-shore was simply not a plausible argument, and probably detrimental to one's own future.
Result? China became the world's factory.
Many shareholders all over the world are terrible at taking a long term view - get rich quick or get outta here. They're to blame, they're a scourge on our society.
OK, who are they? Ah, it's our pension schemes. That means practically all of us are ultimately to blame.
The problem with Trump's strategy (let's call it that for the moment) is that imposing trade barriers effectively forces there to be a great reckoning. Just how much money does America have? How well has America invested in the education and training of its own people? How willing are Americans (remember that he's kicking out the illegal immigrants too) to do boring, low paid manufacturing work? Are they prepared to accept that a made-in-America microwave oven of any sort really should cost $400, not $50? None of these questions have good answers...
...That a lie can run round the world before the truth has got its boots on.
The difference today is that the lie gets a helpful push from the likes of Google and Facebook who then glue the truth's boots to the starting blocks.
CERN's data shows how wobbly the planet is, really. They've previously reported correlation between beam position and waves crashing ashore in the Bay of Biscay. So if nothing else they can give worthwhile surf reports for the Atlantic coast of Europe!
Speaking as an RF engineer, it sounds perfectly plausible to me. Though I suspect they need the phone to be relatively still for it to work.
Another way to defeat it is to not use public WiFi.
And it doesn't force them to install any of their apps either. As proven by Amazon, hardware makers have the right to default android and avoid installing any of Google apps.
You're forgetting the role of Google Play Services. Android app writers are forced to write their apps to use these services if the want access to things like mapping, messaging, etc on mainstream Android. Google do not open source Google Play Services. So they're not on Amazon's version of Android.
So App developers are limited in what their app can achieve if they want to target both Google's version of Android and Amazon's without having two separate code bases. So guess what; the app developers target their software for Google's version of Android with Play Services, because that had the biggest market share.
And because Apps are reliant on Google Play Services, the phone manufacturers have to succumb to Google's terms and conditions if they want to make a phone that is marketable.
This strategy has worked very well for Google, but it backfired in China. Baidu was big enough and quick enough to be able to displace Google's Play Services with their own equivalent. There's 1billion+ Android users in China, non of whom are paying into Google's coffers by using Google's services. It's backfiring for Apple too - incompatibility with Baidu's services is becoming a market killer in China, no matter how shiny your phone is. In many ways Baidu have outstripped Google's services, for example offering payment systems that work widely long before Google Pay.
Some OSes like VxWorks have long had nice facilities for post-mortem debugging. With that you can write your software to log various things in a circular buffer, as well as a ton of OS / task runtime scheduling data. If it ever goes bang you simply connect the debug environment (or extract the buffer content somehow) and see what had happened in the time leading up to the crash. Very nice. And being a nice, tidy, tiny OS makes it well suited to an IoT device.
And you can achieve similar results with Linux these days; stuff like FTRACE and kernelshark spring to mind. Maybe Linux needs more resources to run than VxWorks, and this matters a lot for price conscious consumer products.
These are all nice things to have, but in microcontroller land we don't really get an OS. We know we want that kind of post mortem capability, but it's a pain to have to develop it all oneself.
Biting the hand that feeds IT © 1998–2017