Sure, but there are many areas of EE where demand has fallen. Programmable logic has drastically reduced the need for boards full of TTL chips. FPGAs, and even many ASICs, are designed with fully synchronous digital logic, that requires zero knowledge of most EE concepts, and can be done by any kid bright enough to master Verilog/VHDL. My company has done several successful FPGA projects, none of which involved anyone with an EE degree. ADCs, DACs, PWM, and DSPs come built into many microcontrollers, which themselves increasingly come on standard PCBs, with free downloadable libraries to handle all the interfacing.
And who do you think designs those things?
EEs are very much in demand, however they're not in demand for the old "computer engineer" or "programmer" style jobs.
Digital logic design still commands a small premium because it involves a LOT of advanced technology.
Though if you want to be in a field that's in resurgence, analog IC design is in. Even in the digital world - high speed digital signals behave in very basic analog ways that if your experience is in HDLs, you're not going to be able to figure out easily. Analog designers can command 6 figures easy, especially as modern PHYs are analog in nature, so you have to do mixed-signal ICs. Just because joe end user doesn't have to worry about it doesn't mean someone doesn't.
Then there's plenty of analog designs out there. A popular field is power engineering - you know, utility scale. Utilities all over the world are hurting because there really are only a handful of graduates in power engineering, not enough to replace the growing crowd that is retiring. (Yes, the flashy nature of computers and technologies have sapped the talent pool for other sub disciplines). Enough so that starting salaries are close to, if not above, 6 figures. Even those who want to retire are often asked to hang on because there's no one to replace them, or pass the institutional knowledge to.
There's plenty of RF work as well - WiFi and the like are easy to use, but that's because the RF guys made it simple enough to do. Even so, goof the design and you'll be wondering why you have limited range.
Personally, I hate fumbling with MicroUSB cables and my phone. I don't exactly have sausage fingers, but trying to put in that cable when I'm half asleep, the light on my nightstand is off (and I've been reading an eBook) and the end of the cable is loose *somewhere* on the nightstand is really annoying,
Other than the loose cable "problem" (which most people solve with a bit of tape, a binder clip, or other mechanism including $10 "solutions" that basically hold the end down), the problem is you're complaining about micro USB, a horrendous connector.
Sorry, this is an Apple article - and the Lightning connector is much nicer to deal with. If you wonder why Apple still made their own connector over using micro USB, think why did the USB Type C connector was invented.
Everyone blasts Apple on their proprietary connectors. Yet there are valid reasons why they exist - just because something is standard doesn't mean it doesn't suck (like micro USB).
It is pretty difficult to get perfect performance out of a cell modem, the underlying theory is pretty complex, and translating these complex algorithms into a practical working implementation is incredibly difficult. Neither Intel nor Mediatek know how to close the gap. Qualcomm is probably the only company in the world that has the knowhow and brainpower to do this.
Actually, Intel probably has plenty of experience making modems - Intel's chips aren't their own designs, they purchased Infineon, one of the big modem chip manufacturers out there - if it wasn't Qualcomm, it's Infineon.
In fact, the first iPhones use Infineon modems. Though, one reason to use Infineon was extreme power management - so much so that it was responsible for killing AT&T's network. Basically the modems immediately killed the data channel when the transfer was over, so if you're websurfing, it would open a bunch of data channels, transfer the content, then shut them down. The end result was an overloaded control channel (opening and closing data channels is a control message). With enough iPhones, this consumed all control channel bandwidth. End result? Dropped calls because a tower with a full control channel mean the phone cannot do a handoff. It's why AT&T, despite having the worst call drop history, had some of the best data transfer rates (because a full control channel doesn't have a relation to how many voice/data channels are available).
As for Intel, maybe Intel is trying to court Apple - with modem chips at first, then maybe with SoC business.
Of course, it could be the LAPD needs to justify the huge expense of patrolling from Ghetto Birds instead of ground-based black-and-whites, and they're not at all bothered by the statistical insignificance of the small sample trotted out here as causation.
Ground patrol might be difficult in LA traffic, I would imagine... and flying LA isn't much fun either (lots of air traffic to contend with) but at least there isn't the chance of running into gridlock.
Uber is just forcing free market economics on governments that don't want it, and surprise surprise, prices plummet while service improves greatly. Get rid of the damn medallions and be done with it.
The problem is the free market sucks for utilities.
Uber works, but it's only working on cherry-picked routes and times. Taxis are heavily regulated not just in the drivers and licensing, but also in what they can do. For example, most taxis are required to pick up drunks and take them home, and dealing with a drunk is not an easy thing (think having to clean up your car afterwards). Likewise, most taxis must pick up their fares regardless of color, creed, or other discriminatory factor. And they have to cover the whole city - they may not want to go into a low-rent district, but if they accept the call they have to.
Uber drivers, though, are free to not do any of those things. If you don't wan to pick up some guy because he's black, just drive along. (In many places, a taxi driver doing this would be forced to call another taxi AND wait for that cab to arrive - they're not allowed to drive off).
Then again, taxi companies are evil. But I suppose it's OK when you find yourself partying on a Friday night and unable to get home because there's no taxis and uber isn't willing to pick up people who might throw up in the vehicle.
This experiment only documents the survivability of the NAND Flash itself, really. I've had two consumer SSDs and at least one SD fail completely for other reasons; they became completely un-usable, not just un-writable. In the case of the SSDs at least, I was told it was due to internal controller failure, meaning the NAND itself was fine but the circuits to control and access it were trashed. I suppose a platter-drive analog to that would be having the platters in mint condition with all data intact but the servo coil melted, or something.
Since I've only owned three consumer SSDs and two of those died from a mode of failure that wasn't even addressed by this experiment, what am I to make of the real value of the results? They certainly have no meaning for me, but YMMV.
Well, this test is to figure out if the lifetime of an SSD is adequate - everyone knows flash life is limited, is it limited to the point where every write you should cringe or will it handle enough data that you needn't worry about it?
In this case, the tests show it's closer to the latter.
As for disastrous failure modes, the most common one is FTL table corruption (flash translation later). The tables map the externally visible sectors to the internal flash (you want to wear level the flash so no one block gets unduly worn out even if you repeatedly write to it).
The problem is those tables in cheap SSDs are cached in RAM for speed, but they don't have much in the way of a backup power supply to dump the cache back into the media. It's possible during a write back that the power fails and the table gets corrupt. On the next boot, it fails to load the tables and it's dead.
Those you can usually save if you do an ATA_SECURE_ERASE command which resets the table mappings back to default.
Or get better SSDs that build in capacitance so they can do emergency write backs.
Actually that means it runs Linux natively, which is kind of a big draw from my perspective. I'm considering getting one, but will not be running ChromeOS on it if I do.
Only if by big draw you like kludges. Sure it may be the Linux way but still.
Yeah, you CAN run Linux on it. You can also run Windows (it has SeaBIOS in it). But to do either means you have hit Ctrl-D within 30 seconds of power up (or reboot) every time to boot into your "alternate" (non-ChromeOS) OS otherwise it times out and goes into a recovery mode where it waits for you to insert a recovery USB stick. Sure not a hassle in that you can turn it off and turn it on again, then wait for it to get to the point where it finds an unsigned OS so you can hit Ctrl-D, but still not elegant.
So yes, you can, but it's not a Linux laptop by far.
The other fatal assumption is that Apple is going to sell 20M watches.
Android and Pebble combined barely tilt at 1M (Samsung's latest generation offerings, barely sold 300K combined).
So now Apple is going to sell not just as many Apple Watches as smartwatches combined, but 20 times as much? Granted, interest is high, and even a product like the iPad was mocked as being something completely without reason but still.
but connected to proprietary products instead of a server that runs on almost any hardware
I never understood why people liked the squeezebox better than other alternatives. I had an AudioTron, and it required zero software installation on Windows and OS X, and one useful package on Linux. It relied solely on SMB and didn't need any indexing server or anything. You gave it a user account and it could either self discover the shares or you could explicitly point your music share to it and it indexed that.
Made using a special server sorta like having to use iTunes to load your MP3 player.
There's also gEDA which is an open (GPL'd) EDA suite including a schematic editor, PCB layout tool, and a bunch of other EDA tools.
The big thing with open hardware is simply getting the hardware - RPi and Arduinos are popular because it's easy to get the hardware for minimal cost, and many people make it on behalf of others (well, not the Pi, but that's because of Broadcom).
Open hardware requires the ability to make money (i.e, commercialize) the design. This is not the evil "we will sell your design to make millions" theme, but more so companies can take Open Hardware designs and build them for you. Or at least assemble you a kit.
Nothing screams "useless" more than seeing an NC label slapped on an open hardware design because it means if I wanted to build it, I have to source it all myself instead of being able to go to some company to get it all kitted up or even assembled and I just click "buy it now".
What the summary said was that the timestamps are consistent with an 8-5 day in those time zones, not that the timestamps came from those timezones. Timestamps aren't UTC anything -- they're milliseconds since epoch (generally), and the OS converts on the fly when displaying. I can't speak for the NSA, but core hours are 10-3 for many government workers, and many people go in to the office early to beat traffic. Also, the NSA is under the DoD, and DoD tends to get an early start. All of that is consistent with what one would expect to see.
And to address the GP, the odds of finding a string that matches a codeword, especially a unique codeword, are very slim. Probably millions to one. You're not going to find, say, "XKEYSCORE" in Microsoft or Apple source code. That's the most convincing evidence -- the timestamp stuff is just icing.
I expect to see future exploits released with standardized timestamps and obfuscated strings.
I find it very circumstantial and more akin to fitting the evidence to the crime. I mean, are the only software developers who work normal business hours on normal workdays in the Eastern timezone all working for the NSA? I find that extremely hard to believe, even more so when you consider that a lot of developers do work on the east coast (sorry, software development is not an exclusively west coast thing).
Even a symbol like "Backsnarf" sounds like something that could plausibly be used in malware to indicate reverse snarfing of whatever it is.
Ditto XKEYCODE. Sounds like something someone might call a keyboard map - either the mapping driver or a keymap.
You can have negative feedback without negativity.
Negativity is a virus that once it infects and takes hold, spreads rapidly and kills productivity and innovation.
Negative feedback though is a positive thing, provided it's done correctly - i.e., it's not negativity, it's constructive criticism. The difference is that negativity focuses on the bad alone, while constructive criticism focuses on the rehabilitation.
"This design is stupid. You're an idiot" is a negative statement that spreads negativity. "This design is stupid because you're not using the new architecture features that are going to be present in the new release and instead trying to reinvent the wheel" is negative feedback that becomes constructive because it now presents a resolution to the problem.
It also turns the feedback giver from someone who always says no to someone who provides helpful assistance.
If all you do is complain and bitch about everyone doing crap for work, one of two things happens - either it infects others and it turns into everyone bitching about everyone else and no work gets done, or you'll find yourself isolated as being difficult to work with. Add in racism/sexism/etc and other offensive comments (which have no place in the modern workplace or anywhere for that matter) and either you're out of a job or no one wants to work with you anymore.
Hell, even Linux goes on rants, but at least he tries to justify his rant by giving feedback on what's wrong. He lacks tact and diplomacy, but at least he clearly explains why it's bad, and he attacks the technical content, not the person.
The most ridiculous thing about the new Macbook is that you can't charge it and use USB at the same time unless you buy a $80 dongle
For now. But given USB Type C has been out over a year now, it's only a matter of time before someone else makes a MacBook to everything adapter.
If the only complaint about the MacBook is that USB connector, I'd say Apple did well - it's a non-proprietary connector that everyone's had a year to release stuff for. And given how few devices are out there, it appears that Apple is again going to forge the production of a pile of new USB devices like it did way back in the late 90s when it went USB only (and the only USB things were overpriced keyboards and mice).
Perhaps blame the USB accessory manufacturers for sitting on USB Type C.
...when they made the memory in the new Mac Minis impossible to upgrade and reduced their performance. The late 2012 quad-core model is still the fastest, best one they ever made.
I agree about the RAM - that is stupid, however, the reason for the dual core is simple - Intel doesn't make the i7 in the required socket formfactor. The i5s and i7s used in the Mac Mini are the same socket, so it's a single design.
So if Apple wanted to offer the i5 and i7, they had to either design two Mac Mini motherboards, one for each, or use the slower i7 because that's all Intel has.
It's not the ONLY time Apple's been hampered by Intel's lineup.
And the Mac Mini isn't exactly Apple's top seller. It joins the Mac Pro in the worst sellers in the lineup. So no, the dual motherboard idea is not flying.
But the soldered RAM on the Mini doesn't make sense - there's no compelling reason for it - there's no space limits (the current Mac Mini is the same volume as the old one) like there is in the portable lineup, and there's nothing in the new mini that takes up so much space that the RAM couldn't be accomodated.
This actually worked! In the early days of film the Library of Congress had each frame of entire films printed onto paper to establish copyright.
That's because in the early days of movies, we didn't have "moves" as a copyrightable item. So movies, being motion pictures, were printed onto paper and copyrighted that way as photographs. It was only later that movies were copyrightable in and of themselves and you didn't have to work around it by printing it to paper.
you know that DRAM hack-attack that was just made public? how much you wanna bet the US gov had a hand in making that possible?
TFA mentions several things. First, they tried to write their own version of Xcode and tools to be able to substitute it on a victim's machine, they also tried to crack Apple's keys (which TFA claims they didn't manage to do) - it's unclear if it's Apple's signing keys, the per-device iOS keys, or what) etc.
I think the CIA would've had an easier time if they just jailbroke the devices. Or given how thoroughly a jailbreak removes iOS security, perhaps they're the ones releasing all those jailbreaks to encourage people to use them?
Of course, I can't remember if iOS devices use encrypted RAM - memory encryption units are common and they are less vulnerable to rowhammer because the bit ordering gets scrambled. It's hard to hit a particular bit if it happens that the key used moves your attack bit to a different bit. And the bit order changes with memory address.
And the key area is often loaded from the hardware RNG on startup so rebooting the device means the memory scrambling function for those rows changes.
Power should be easy - electricity has been around for a couple of centuries, so all you really need to do is provide a break out cable and say "Ground" and "+5V @ 1A". (We've used volts forever). This is especially easy since volts and amps are based on fundamental constants so even if in 50 years they went to dabblequads and quibblewhats, it's a trivial matter to convert between the two units.
And yeah, ye olde analog media is best. Film or even printed paper can be easily preserved, and it's really easy to restore if it deteriorates.
This media is somewhat easy to handle and restore, so as long as human intelligence doesn't dwindle over the next 40 years, it should be fine.
If you want to store it digitally, don't use any encoded format - you'ld basically want to store it in something like the equivalent of film - as uncompressed bitmaps stored separately in sequentially numbered files. So if they're able to read the data off it, each frame is by itself (so deterioration doesn't disrupt the frame boundaries) and converting RGB to whatever their display technology is would be a trivial exercise. The problem with encoded formats is data loss - they're robust in that they use sync bytes to regain sync, but you can lose a lot of video data simply by losing the wrong bits. Also, encoded formats, if they're way obsolete, will be very difficult to decode - imagine pulling it out and now someone has to go and write a codec from the documentation you included.
A simple frame-based bitmap means each frame can be individually decoded, the documentation is simple and you can look at it with a hex editor and see if it "looks right". And that's after decoding enough to get at the data on the disc or hard drive.
Especially compared to those damn 4 dimensional USB connectors: try to plug it in, fail, reverse, fail *again*, reverse once more, *then* it will go in.
You can consider USB ports to be a spin-1/2 device - it takes 2 complete rotations for a USB connector to return to its original state. So you try it once, fail, flip it over, try it again, fail, flip it over. Note that even though you did a 360 rotation of the connector, the connector's not in the same state it was 360 degrees ago - it's still only part way through its spin. You have to rotate it another 360 degrees to return to its original state.
Yes, who knew the USB Forum guys were quantum mechanics wizards.
I have 2 generic servers in my closet that use Thunderbolt to talk to big ass arrays of disks. Nothing Apple related about them.
And USB 3 does not do everything I use Thunderbolt for on my Mac, including ferry USB3 over the same wire as video. I come home (or go to the office) and plug in my laptop with single cable and instantly my displays, USB3 devices, audio and networking all work... without eating a ridiculous amount of CPU power as required by USB.
Dear god, do not drop Thunderbolt support based on the silly musings of a bunch of people buying the cheapest crap hardware they can possibly buy and then being pissy they don't have the same functionality. Fortunately Apple doesn't generally listen to a bunch of whiners on slashdot.
Thunderbolt has several advantages over DockPort.
First, it's effectively PCIe - that should already start brewing ideas. Instead of crappy USB-to-serial adapters or parallel adapters that barely work, a Thunderbolt variant would work just like a real connector on your PC (and is practically driverless).
Thunderbolt also has the uncanny ability to hook up huge daisy chains of drives without losing too much speed between the first and last drive - most of the loss in speed comes from having more devices on the line than the actual order of them. If you want to deal with big ass external arrays, Thunderbolt makes that all the more convenient.
Heck, USB generally sucks for storage until you find a matching pair of UAS (USB Attached SCSI) host controllers and drives (which are $$$).
Except that the ECC memory only costs so much because so few people buy it. It's a "business part". I don't think most consumer mobos are equipped to handle ECC memory either. It's a shame too because if the costs were in line with the actual hardware (it cost $112 instead of $160) and it was supported by the mobo manufacturers then I think a lot of system builders would go for ECC memory.
It's chipset and processor support, actually.
Intel, for example, typically mandates ECC on the Xeon line (modern CPUs have the memory controller on die now, so ECC is dictated by the processor chosen).
It's also available on the workstation line of processors since workstations typically go for stability.
Of course, ECC is still a "feature" people will pay an inflated markup for so those Xeons and workstation processors cost a fair bit more than the low end and enthusiast parts.
(Enthusiast parts? Yes, those don't typically include ECC as ECC typically ads a clock or two of delay, and people wanting "extreme performance" are trying get the fastest RAM possible - ECC adding a clock or two (with buffering/registers) just defeats the whole purpose of buying fast RAM). Look it up some time on how going from CL6 to CL5 will cost you extra.
Take a look at Iridium or GlobalStar, the only two Low Earth Orbit satellite phone companies I know of. How come do they keep on finding new investors when they have such a poor track record at making money?
Because their biggest customer is governments. GlobalStar and Iridium are heavily used by the military (all countries) in order for communications support at practically anywhere in the world. Then there are plenty of countries where they have huge swaths of barren land but you still need to have people do work there.
Basically satellite phones have become essential technology in the backwoods where there is no cell coverage - and when your life is on the line, $1/minute emergency call is cheap.
I think the US DoD actually bought Iridium back when they went bankrupt in the early 00's, so it's owned and operated by the DoD.
Training dogs is expensive, especially if you're not sure of the results. Will a dog be able to sniff for the biomarkers for cancer? There's anecdotal evidence prior that maybe they could (untrained dogs have been shown to detect cancers in their owners).
But to train one specially to detect the cancer can be quite expensive (most service dogs cost around $30,000 and 2 years to train from puppyhood, and that excludes the cost of those found unsuitable partway through. Knowing breeders who do donate to service dog organizations, they try to select those who they feel make the best candidates, but sometimes it just doesn't work out. Of course, these are reputable breeders and it turns out when you do all this, there's not actually a lot of money left over).
Then you need to figure out which breed might make the best detector dogs - the time and cost it takes to raise just one dog for the study means by the time you get around to trying to reproduce it, it's going to be several more years. (In the meantime, said study dog will be put to work and studied more to see how effective it is over time).
Doesn't need it, since Android's init system does basically all of systemd without being GPL. Using systemd would move Android backwards since the whole goal is to have practically everything be Apache licensed.
Manages services, triggers (you can run things based on events), handles logging, properties, IPC, udev, etc. All in one handy executable, too.
I can't see how it would be possible to defeat ECC.
The attacker would have to construct a write that affects the desired bits in the row-to-be-hammered and has check bits that affect the row-to-be-hammered's check bits such that the altered row is validated. This is probably nigh impossible to do in all but a select few constrained cases.
It's possible, but very unlikely.
Rowhammer is not new - it's been known since the 90s since it affects NAND flash memory as well (the same stuff in an SSD) - here there are two problems. It's called write disturb and read disturb. Because in NAND flash, all the storage transistors are wired in series - each transistor is a page of flash, and all the pages are wired in series, so reading one page requires the transistors in other pages to be activated to be able to read the desired page.
So NAND flash manufacturers design their transistors sepecially so you minimize write disturbs (where writing to a page flips bits in other pages) as long as you write the pages in sequence. But there's also read disturbs where the act of reading a page can flip bits on other pages because you're still activating those series transistors.
That's why NAND flash has the spare area after each page - it's for ECC data which is required to catch such errors.
Slashdot Mirror
And who do you think designs those things?
EEs are very much in demand, however they're not in demand for the old "computer engineer" or "programmer" style jobs.
Digital logic design still commands a small premium because it involves a LOT of advanced technology.
Though if you want to be in a field that's in resurgence, analog IC design is in. Even in the digital world - high speed digital signals behave in very basic analog ways that if your experience is in HDLs, you're not going to be able to figure out easily. Analog designers can command 6 figures easy, especially as modern PHYs are analog in nature, so you have to do mixed-signal ICs. Just because joe end user doesn't have to worry about it doesn't mean someone doesn't.
Then there's plenty of analog designs out there. A popular field is power engineering - you know, utility scale. Utilities all over the world are hurting because there really are only a handful of graduates in power engineering, not enough to replace the growing crowd that is retiring. (Yes, the flashy nature of computers and technologies have sapped the talent pool for other sub disciplines). Enough so that starting salaries are close to, if not above, 6 figures. Even those who want to retire are often asked to hang on because there's no one to replace them, or pass the institutional knowledge to.
There's plenty of RF work as well - WiFi and the like are easy to use, but that's because the RF guys made it simple enough to do. Even so, goof the design and you'll be wondering why you have limited range.
Other than the loose cable "problem" (which most people solve with a bit of tape, a binder clip, or other mechanism including $10 "solutions" that basically hold the end down), the problem is you're complaining about micro USB, a horrendous connector.
Sorry, this is an Apple article - and the Lightning connector is much nicer to deal with. If you wonder why Apple still made their own connector over using micro USB, think why did the USB Type C connector was invented.
Everyone blasts Apple on their proprietary connectors. Yet there are valid reasons why they exist - just because something is standard doesn't mean it doesn't suck (like micro USB).
Actually, Intel probably has plenty of experience making modems - Intel's chips aren't their own designs, they purchased Infineon, one of the big modem chip manufacturers out there - if it wasn't Qualcomm, it's Infineon.
In fact, the first iPhones use Infineon modems. Though, one reason to use Infineon was extreme power management - so much so that it was responsible for killing AT&T's network. Basically the modems immediately killed the data channel when the transfer was over, so if you're websurfing, it would open a bunch of data channels, transfer the content, then shut them down. The end result was an overloaded control channel (opening and closing data channels is a control message). With enough iPhones, this consumed all control channel bandwidth. End result? Dropped calls because a tower with a full control channel mean the phone cannot do a handoff. It's why AT&T, despite having the worst call drop history, had some of the best data transfer rates (because a full control channel doesn't have a relation to how many voice/data channels are available).
As for Intel, maybe Intel is trying to court Apple - with modem chips at first, then maybe with SoC business.
Ground patrol might be difficult in LA traffic, I would imagine... and flying LA isn't much fun either (lots of air traffic to contend with) but at least there isn't the chance of running into gridlock.
The problem is the free market sucks for utilities.
Uber works, but it's only working on cherry-picked routes and times. Taxis are heavily regulated not just in the drivers and licensing, but also in what they can do. For example, most taxis are required to pick up drunks and take them home, and dealing with a drunk is not an easy thing (think having to clean up your car afterwards). Likewise, most taxis must pick up their fares regardless of color, creed, or other discriminatory factor. And they have to cover the whole city - they may not want to go into a low-rent district, but if they accept the call they have to.
Uber drivers, though, are free to not do any of those things. If you don't wan to pick up some guy because he's black, just drive along. (In many places, a taxi driver doing this would be forced to call another taxi AND wait for that cab to arrive - they're not allowed to drive off).
Then again, taxi companies are evil. But I suppose it's OK when you find yourself partying on a Friday night and unable to get home because there's no taxis and uber isn't willing to pick up people who might throw up in the vehicle.
Well, this test is to figure out if the lifetime of an SSD is adequate - everyone knows flash life is limited, is it limited to the point where every write you should cringe or will it handle enough data that you needn't worry about it?
In this case, the tests show it's closer to the latter.
As for disastrous failure modes, the most common one is FTL table corruption (flash translation later). The tables map the externally visible sectors to the internal flash (you want to wear level the flash so no one block gets unduly worn out even if you repeatedly write to it).
The problem is those tables in cheap SSDs are cached in RAM for speed, but they don't have much in the way of a backup power supply to dump the cache back into the media. It's possible during a write back that the power fails and the table gets corrupt. On the next boot, it fails to load the tables and it's dead.
Those you can usually save if you do an ATA_SECURE_ERASE command which resets the table mappings back to default.
Or get better SSDs that build in capacitance so they can do emergency write backs.
Only if by big draw you like kludges. Sure it may be the Linux way but still.
Yeah, you CAN run Linux on it. You can also run Windows (it has SeaBIOS in it). But to do either means you have hit Ctrl-D within 30 seconds of power up (or reboot) every time to boot into your "alternate" (non-ChromeOS) OS otherwise it times out and goes into a recovery mode where it waits for you to insert a recovery USB stick. Sure not a hassle in that you can turn it off and turn it on again, then wait for it to get to the point where it finds an unsigned OS so you can hit Ctrl-D, but still not elegant.
So yes, you can, but it's not a Linux laptop by far.
The other fatal assumption is that Apple is going to sell 20M watches.
Android and Pebble combined barely tilt at 1M (Samsung's latest generation offerings, barely sold 300K combined).
So now Apple is going to sell not just as many Apple Watches as smartwatches combined, but 20 times as much? Granted, interest is high, and even a product like the iPad was mocked as being something completely without reason but still.
No, like many surprised iPhone users, he discovered that extra smurfberries are NOT free, and someone has to pay the bill.
It's just unlike Apple, Facebook wasn't so willing to refund all that accidentally-spent money.
I never understood why people liked the squeezebox better than other alternatives. I had an AudioTron, and it required zero software installation on Windows and OS X, and one useful package on Linux. It relied solely on SMB and didn't need any indexing server or anything. You gave it a user account and it could either self discover the shares or you could explicitly point your music share to it and it indexed that.
Made using a special server sorta like having to use iTunes to load your MP3 player.
There's also gEDA which is an open (GPL'd) EDA suite including a schematic editor, PCB layout tool, and a bunch of other EDA tools.
The big thing with open hardware is simply getting the hardware - RPi and Arduinos are popular because it's easy to get the hardware for minimal cost, and many people make it on behalf of others (well, not the Pi, but that's because of Broadcom).
Open hardware requires the ability to make money (i.e, commercialize) the design. This is not the evil "we will sell your design to make millions" theme, but more so companies can take Open Hardware designs and build them for you. Or at least assemble you a kit.
Nothing screams "useless" more than seeing an NC label slapped on an open hardware design because it means if I wanted to build it, I have to source it all myself instead of being able to go to some company to get it all kitted up or even assembled and I just click "buy it now".
I find it very circumstantial and more akin to fitting the evidence to the crime. I mean, are the only software developers who work normal business hours on normal workdays in the Eastern timezone all working for the NSA? I find that extremely hard to believe, even more so when you consider that a lot of developers do work on the east coast (sorry, software development is not an exclusively west coast thing).
Even a symbol like "Backsnarf" sounds like something that could plausibly be used in malware to indicate reverse snarfing of whatever it is.
Ditto XKEYCODE. Sounds like something someone might call a keyboard map - either the mapping driver or a keymap.
You can have negative feedback without negativity.
Negativity is a virus that once it infects and takes hold, spreads rapidly and kills productivity and innovation.
Negative feedback though is a positive thing, provided it's done correctly - i.e., it's not negativity, it's constructive criticism. The difference is that negativity focuses on the bad alone, while constructive criticism focuses on the rehabilitation.
"This design is stupid. You're an idiot" is a negative statement that spreads negativity. "This design is stupid because you're not using the new architecture features that are going to be present in the new release and instead trying to reinvent the wheel" is negative feedback that becomes constructive because it now presents a resolution to the problem.
It also turns the feedback giver from someone who always says no to someone who provides helpful assistance.
If all you do is complain and bitch about everyone doing crap for work, one of two things happens - either it infects others and it turns into everyone bitching about everyone else and no work gets done, or you'll find yourself isolated as being difficult to work with. Add in racism/sexism/etc and other offensive comments (which have no place in the modern workplace or anywhere for that matter) and either you're out of a job or no one wants to work with you anymore.
Hell, even Linux goes on rants, but at least he tries to justify his rant by giving feedback on what's wrong. He lacks tact and diplomacy, but at least he clearly explains why it's bad, and he attacks the technical content, not the person.
For now. But given USB Type C has been out over a year now, it's only a matter of time before someone else makes a MacBook to everything adapter.
If the only complaint about the MacBook is that USB connector, I'd say Apple did well - it's a non-proprietary connector that everyone's had a year to release stuff for. And given how few devices are out there, it appears that Apple is again going to forge the production of a pile of new USB devices like it did way back in the late 90s when it went USB only (and the only USB things were overpriced keyboards and mice).
Perhaps blame the USB accessory manufacturers for sitting on USB Type C.
I agree about the RAM - that is stupid, however, the reason for the dual core is simple - Intel doesn't make the i7 in the required socket formfactor. The i5s and i7s used in the Mac Mini are the same socket, so it's a single design.
So if Apple wanted to offer the i5 and i7, they had to either design two Mac Mini motherboards, one for each, or use the slower i7 because that's all Intel has.
It's not the ONLY time Apple's been hampered by Intel's lineup.
And the Mac Mini isn't exactly Apple's top seller. It joins the Mac Pro in the worst sellers in the lineup. So no, the dual motherboard idea is not flying.
But the soldered RAM on the Mini doesn't make sense - there's no compelling reason for it - there's no space limits (the current Mac Mini is the same volume as the old one) like there is in the portable lineup, and there's nothing in the new mini that takes up so much space that the RAM couldn't be accomodated.
That's because in the early days of movies, we didn't have "moves" as a copyrightable item. So movies, being motion pictures, were printed onto paper and copyrighted that way as photographs. It was only later that movies were copyrightable in and of themselves and you didn't have to work around it by printing it to paper.
TFA mentions several things. First, they tried to write their own version of Xcode and tools to be able to substitute it on a victim's machine, they also tried to crack Apple's keys (which TFA claims they didn't manage to do) - it's unclear if it's Apple's signing keys, the per-device iOS keys, or what) etc.
I think the CIA would've had an easier time if they just jailbroke the devices. Or given how thoroughly a jailbreak removes iOS security, perhaps they're the ones releasing all those jailbreaks to encourage people to use them?
Of course, I can't remember if iOS devices use encrypted RAM - memory encryption units are common and they are less vulnerable to rowhammer because the bit ordering gets scrambled. It's hard to hit a particular bit if it happens that the key used moves your attack bit to a different bit. And the bit order changes with memory address.
And the key area is often loaded from the hardware RNG on startup so rebooting the device means the memory scrambling function for those rows changes.
Power should be easy - electricity has been around for a couple of centuries, so all you really need to do is provide a break out cable and say "Ground" and "+5V @ 1A". (We've used volts forever). This is especially easy since volts and amps are based on fundamental constants so even if in 50 years they went to dabblequads and quibblewhats, it's a trivial matter to convert between the two units.
And yeah, ye olde analog media is best. Film or even printed paper can be easily preserved, and it's really easy to restore if it deteriorates.
This media is somewhat easy to handle and restore, so as long as human intelligence doesn't dwindle over the next 40 years, it should be fine.
If you want to store it digitally, don't use any encoded format - you'ld basically want to store it in something like the equivalent of film - as uncompressed bitmaps stored separately in sequentially numbered files. So if they're able to read the data off it, each frame is by itself (so deterioration doesn't disrupt the frame boundaries) and converting RGB to whatever their display technology is would be a trivial exercise. The problem with encoded formats is data loss - they're robust in that they use sync bytes to regain sync, but you can lose a lot of video data simply by losing the wrong bits. Also, encoded formats, if they're way obsolete, will be very difficult to decode - imagine pulling it out and now someone has to go and write a codec from the documentation you included.
A simple frame-based bitmap means each frame can be individually decoded, the documentation is simple and you can look at it with a hex editor and see if it "looks right". And that's after decoding enough to get at the data on the disc or hard drive.
You can consider USB ports to be a spin-1/2 device - it takes 2 complete rotations for a USB connector to return to its original state. So you try it once, fail, flip it over, try it again, fail, flip it over. Note that even though you did a 360 rotation of the connector, the connector's not in the same state it was 360 degrees ago - it's still only part way through its spin. You have to rotate it another 360 degrees to return to its original state.
Yes, who knew the USB Forum guys were quantum mechanics wizards.
Thunderbolt has several advantages over DockPort.
First, it's effectively PCIe - that should already start brewing ideas. Instead of crappy USB-to-serial adapters or parallel adapters that barely work, a Thunderbolt variant would work just like a real connector on your PC (and is practically driverless).
Thunderbolt also has the uncanny ability to hook up huge daisy chains of drives without losing too much speed between the first and last drive - most of the loss in speed comes from having more devices on the line than the actual order of them. If you want to deal with big ass external arrays, Thunderbolt makes that all the more convenient.
Heck, USB generally sucks for storage until you find a matching pair of UAS (USB Attached SCSI) host controllers and drives (which are $$$).
It's chipset and processor support, actually.
Intel, for example, typically mandates ECC on the Xeon line (modern CPUs have the memory controller on die now, so ECC is dictated by the processor chosen).
It's also available on the workstation line of processors since workstations typically go for stability.
Of course, ECC is still a "feature" people will pay an inflated markup for so those Xeons and workstation processors cost a fair bit more than the low end and enthusiast parts.
(Enthusiast parts? Yes, those don't typically include ECC as ECC typically ads a clock or two of delay, and people wanting "extreme performance" are trying get the fastest RAM possible - ECC adding a clock or two (with buffering/registers) just defeats the whole purpose of buying fast RAM). Look it up some time on how going from CL6 to CL5 will cost you extra.
Because their biggest customer is governments. GlobalStar and Iridium are heavily used by the military (all countries) in order for communications support at practically anywhere in the world. Then there are plenty of countries where they have huge swaths of barren land but you still need to have people do work there.
Basically satellite phones have become essential technology in the backwoods where there is no cell coverage - and when your life is on the line, $1/minute emergency call is cheap.
I think the US DoD actually bought Iridium back when they went bankrupt in the early 00's, so it's owned and operated by the DoD.
Training dogs is expensive, especially if you're not sure of the results. Will a dog be able to sniff for the biomarkers for cancer? There's anecdotal evidence prior that maybe they could (untrained dogs have been shown to detect cancers in their owners).
But to train one specially to detect the cancer can be quite expensive (most service dogs cost around $30,000 and 2 years to train from puppyhood, and that excludes the cost of those found unsuitable partway through. Knowing breeders who do donate to service dog organizations, they try to select those who they feel make the best candidates, but sometimes it just doesn't work out. Of course, these are reputable breeders and it turns out when you do all this, there's not actually a lot of money left over).
Then you need to figure out which breed might make the best detector dogs - the time and cost it takes to raise just one dog for the study means by the time you get around to trying to reproduce it, it's going to be several more years. (In the meantime, said study dog will be put to work and studied more to see how effective it is over time).
Doesn't need it, since Android's init system does basically all of systemd without being GPL. Using systemd would move Android backwards since the whole goal is to have practically everything be Apache licensed.
Manages services, triggers (you can run things based on events), handles logging, properties, IPC, udev, etc. All in one handy executable, too.
It's the "freer" version of systemd.
It's possible, but very unlikely.
Rowhammer is not new - it's been known since the 90s since it affects NAND flash memory as well (the same stuff in an SSD) - here there are two problems. It's called write disturb and read disturb. Because in NAND flash, all the storage transistors are wired in series - each transistor is a page of flash, and all the pages are wired in series, so reading one page requires the transistors in other pages to be activated to be able to read the desired page.
So NAND flash manufacturers design their transistors sepecially so you minimize write disturbs (where writing to a page flips bits in other pages) as long as you write the pages in sequence. But there's also read disturbs where the act of reading a page can flip bits on other pages because you're still activating those series transistors.
That's why NAND flash has the spare area after each page - it's for ECC data which is required to catch such errors.