Patents were already established by the industrial revolution (indeed Watt's extension of steam engine patents was the first documented serious abuse of the patent system). Anything before that is pretty much irrelevant to an industrial economy.
I'm wondering why Edge isn't. Not only would more privacy features be a good differentiator, anything that makes ads less effective would harm Google, which seems like it would be in Microsoft's best interests.
The UK has a lot of loopholes for this. For example, if you're a big chain, you don't have to have sold it at the full price in the store that offers a discount, so they'll often have one store each week selling something at the full price but all of the others selling it at the sale price. You can also go straight from an introductory discount to a sale price with basically no time in the middle when it's full price.
How quick does the key exchange have to be? Most of the interactions with these devices are for non-emergency diagnostics. It's not like you're in an ambulance and the paramedic needs to log into your pacemaker to restart your heart. If the key exchange takes a few minutes, that's fine.
I agree with your heading, but not with the rest of your post. The problem is that the FDA requires that the company have the software certified as safe by a third party, but places very few rules on what this entails. In a lot of cases, the people certifying the software don't even have access to the code: they read the design docs, but nothing else. There's no red teaming of medical device software before widespread deployment and no auditing by the FDA. The FDA is happy to certify such devices as 'safe' with nothing like enough information to be able to honestly make that claim.
Anyone who manages to get Linux running on a pacemaker deserves a lot of credit. These devices consume an order of magnitude or two less power than the smallest systems to run Linux.
I'd have thought it'd be trivial to interface the open source version of Swift with GNUstep
Unlike Clang, the Swift compiler doesn't have very clean interfaces for abstracting the different Objective-C implementations (clang does because I wrote them, before Apple added support for their runtime). Instead, when you build Swift, you have a bunch of #ifdef __APPLE__ things that turn off the Objective-C interop for non-Apple platforms. This also means some quite big ABI changes, for example the Swift CoreFoundation implementation has different object layouts to the ones that Apple uses and the way that casts are handled in Swift is different if Swift objects are allowed to secretly be Objective-C objects.
Well, at least they stopped desecrating Objective-C with Java syntax.
Since Steve Naroff left, I don't think anyone on Apple's toolchain team has really understood what made Objective-C good: a simple and clear object model, orthogonal syntax, and good C/C++ interop.
IMHO, although it's sad they completely removed the dynamic aspect of ObjC, swift is the best of all the toy languages. I was comfortable using it to write a full program right away and I do mostly embedded C. The only downside is that it is a bit of an NP-hard problem to decide whether a correct program will compile after adding a line break.
The hard problems in designing a modern programming language are concurrency, security, and error handling. Swift didn't even really try to address any of these (they've now added a few approaches to error handling, but it needed to be baked into the language and consistently applied throughout the standard library from day one). As such, I don't really see it as a language for the 21st century. It gives me similar abstractions to C++ (which, since C++11, has gone from being a horrible language to being a tolerable one), yet with less widespread support and worse performance.
You might want to read up on the actual history a bit more. Monopoly is based on a came called The Landlord's Game, which has two modes of play. In one, you won by constructing a monopoly, in the other you won by increasing the total size of the economy. The point was to illustrate how unconstrained capitalism would lead to monopolies and negative outcomes for most participants.
The modern version is a set of incremental changes to the 1933 game by Parker (later bought by Hasbro). This was a simplified version of The Landlord's Game, which eliminated the cooperative mode and left outright competition as the only objective.
GNUstep has an implementation of the core GUI frameworks (AppKit and a few others), but the open source version of Swift doesn't have Objective-C interop, which is really the only compelling feature for Swift: it's a good incremental language if you have a lot of legacy Objective-C code.
The thought is that embedded wifi will only ever be a security risk, and could never possibly be of use to anyone.
No, the thought is that embedded WiFi is a security risk and this risk outweighs the benefits. A typical fridge lasts, what, 10-20 years? Do you think LG is going to be back-porting network stack security fixes to Linux for 20 years? Do you think that, even if they wanted to, they will make enough profit on fridges to be able to afford to? Over the last 3-4 years, I've lost track of the number of vulnerabilities that enable anyone who can send a packet to the stack to gain kernel-level privilege. Will LG be fixing all of these for the lifetime of the fridge?
Crowding around it and buying it are two different things. I just bought a new fridge and the most important aspects were size, noise, power consumption (running costs) and price. If I'd seen one with an LCD in a shop, I'd have gone and poked it for novelty value, but it wouldn't have contributed to my purchasing decision (except if it came with any kind of network connectivity to check that it came with a 20 year support contract for security updates).
Steve Jobs' Apple presentations were always a little bit too slick for my taste. You can find him introducing various NeXT technologies on YouTube though, and those are well worth watching. The same style that you'll recognise from the Apple presentations, just not yet quite as polished. Also, not presenting mainstream consumer products, so much more interesting (of very dated) tech.
As the other poster says, the market was changing and the iPhone managed to have perfect timing. A few things about the original iPhone:
No 3G support. I had been using 3G for over a year when the iPhone came out and only supporting EDGE (which networks had to roll out specially for the iPhone) was a bit crappy.
No decent Bluetooth support. I had a Mac when the iPhone came out and I could send SMS and dial my phone directly from the Address Book and get on-screen notifications when my phone rang with my Nokia phone. iPhone users couldn't. It couldn't even send or receive contacts via Bluetooth, as I recall.
No third-party app support. All of the other smartphone vendors supported apps from third parties - it was one of the big things that differentiated smartphones from feature phones. I could even run a port of Doom on my old Nokia phone.
No support for MMS. Admittedly, MMS was a crappy invention that deserved to die horribly, but kids seemed to like it.
No SIP support. My old Nokia phone had integrated support for SIP, so I could make cheap calls via WiFi when available. The iPhone didn't.
It did have a big capacitive touchscreen, but it wasn't even the first phone to do that (though it was a very close-run thing). The iPhone didn't really become interesting until the second version, when they had third-party apps and 3G. The first version was a high-end feature phone that appealed to people who had never owned a smartphone.
So, in this CS degree, did you learn any actual computer science, or did you just pick up specific technologies that will be obsolete in a decade? From that list of things, it sounds like you got a software engineering qualification from a trade school, not a computer science degree.
The important question to ask is how much will it harm the third party if you go out of business. For pretty much anyone using Google services, the answer is 'not at all'.
You do need a microphone, but most people carry a mobile phone and don't look too carefully at the list of apps that are allowed access to the microphone, GPS and WiFi location data, and the network. If you are a state-level adversary, then compromising, say, the Facebook app will let you get this information in a lot of places.
Apple thought people would want that behaviour because that's the behaviour from MacOS 1-9 and people complained a lot when they broke it in 10.0 (and several times later). Now that the die-hard Mac Classic fans are gone, there's less of a reason to care.
That's because the icon view in the Finder is a spacial view. There is no implicit ordering (and it will remember where the icons are in the window), so selecting two files and expecting it to select a range doesn't make sense. I'm slightly surprised that anyone uses a Finder mode other than the NeXT-like browser mode - the spacial view is for people coming from MacOS 9 and hasn't scaled well to modern filesystem sizes for over a decade.
I agree that this sucks, but the fault is with Intel, not Apple. The only quad-core mobile chips that Intel is currently shipping support a maximum of 16GB if you use DDR3 or LPDDR3, or 32GB if you use DDR4 (and don't support LPDDR4). The difference in power between 16GB of LPDDR3 and 32GB of DDR4 is huge and would take 2-3 hours from the battery life (and have a big impact even when in suspend mode, because the RAM remains powered unless you suspend to disk and pay a wake-up time penalty). The next revision ought to support 32GB of LPDDR4, which should be a nice improvement.
The first new PC I bought was an 8086. Over the years parts got upgraded, including motherboards on occasion. A motherboard is a component upgrade, not a whole new PC, so my statement stands
My first PC was an 8086, which I upgraded with a NEC V30H, but no part of that system other than the ISA bus was standard and so it was basically impossible to upgrade beyond sticking a bit more RAM in. Motherboard form factors weren't standardised until the 486, so even keeping the case was usually impossible, and most 486s didn't come with sufficiently powerful power supplies to be upgraded to pentium.
A motherboard is a component upgrade, not a whole new PC, so my statement stands.
It is effectively a whole new PC is the new motherboard requires a new CPU, GPU, and RAM.
Most non-shit laptops have upgradeable CPUs and GPUs
Another variant of this attack used several other mechanisms for generating the sound. If you're doing a very targeted attack, spiking the CPU to 100% until the fans come on and then letting the machine cool gives you a good idea who it is. For a lot of machines, various different operation sequences can make some components emit high-frequency sound that a reasonable microphone can pick up. There was a really neat attack on Tor in data centres about a decade ago that monitored the ambient temperature (using a co-located box's own temperature monitors) to correlate Tor traffic from a particular node with warming of the room, so that (after a lot of samples) you could tell if traffic for a particular user was flowing through the data centre that you were looking at.
That's funny, because when I used to use a desktop PC, I upgraded it piece by piece, never a full PC at once. Worked fine for decades
I find the 'for decades' part of this hard to believe. The first new PC I bought was a 166MHz Pentium clone. Upgrading the CPU required a new motherboard, because newer chips ran at a different voltage. The new motherboard needed PC100 RAM, not PC66, so I also needed to upgrade the RAM. I kept the case, graphics card, hard disk, CD drive, and floppy, but it was effectively a new computer. The next upgrade required a new case (because you couldn't buy AT form-factor motherboards anymore) for the new motherboard. Again, the new CPU needed new RAM, so I was keeping the drives and GPU, but basically nothing else - I also had to replace my ISA NIC with a PCI one, because there were no ISA slots on the new motherboard. The motherboard upgrade after that kept the case, but the drives needed replacing (no IDE on the new motherboard, only SATA), as did the GPU (no AGP, only PCIe).
The times when I did keep something from the old machine, aside from the case that thing quickly ended up being the bottleneck and was replaced shortly after. Eventually, I realised it was cheaper and less effort to just buy a new machine and demote the old one to some other purpose (often meaning giving it to someone else). The only reason for building my own machine now is when I want something that is too niche for any vendor to sell (e.g. the NAS / media centre box that I have running FreeBSD). For a mobile phone, the niche thing that I want is a minimum of 3-5 years of timely security updates for an Android phone and being able to build the hardware myself wouldn't really help there.
Hell, my current laptop has upgraded RAM modules, disk drive, CPU and GPU.
What laptop has an upgradeable CPU and GPU? The only ones I've seen that don't have both soldered onto the board qualify more as luggables than laptops.
Slashdot Mirror
Patents were already established by the industrial revolution (indeed Watt's extension of steam engine patents was the first documented serious abuse of the patent system). Anything before that is pretty much irrelevant to an industrial economy.
I'm wondering why Edge isn't. Not only would more privacy features be a good differentiator, anything that makes ads less effective would harm Google, which seems like it would be in Microsoft's best interests.
The UK has a lot of loopholes for this. For example, if you're a big chain, you don't have to have sold it at the full price in the store that offers a discount, so they'll often have one store each week selling something at the full price but all of the others selling it at the sale price. You can also go straight from an introductory discount to a sale price with basically no time in the middle when it's full price.
How quick does the key exchange have to be? Most of the interactions with these devices are for non-emergency diagnostics. It's not like you're in an ambulance and the paramedic needs to log into your pacemaker to restart your heart. If the key exchange takes a few minutes, that's fine.
I agree with your heading, but not with the rest of your post. The problem is that the FDA requires that the company have the software certified as safe by a third party, but places very few rules on what this entails. In a lot of cases, the people certifying the software don't even have access to the code: they read the design docs, but nothing else. There's no red teaming of medical device software before widespread deployment and no auditing by the FDA. The FDA is happy to certify such devices as 'safe' with nothing like enough information to be able to honestly make that claim.
Anyone who manages to get Linux running on a pacemaker deserves a lot of credit. These devices consume an order of magnitude or two less power than the smallest systems to run Linux.
Inconsistent colour, apparent size varies year to year and even month to month. Seems pretty buggy to me.
I think we should probably think about getting a new one some time about now.
I'd have thought it'd be trivial to interface the open source version of Swift with GNUstep
Unlike Clang, the Swift compiler doesn't have very clean interfaces for abstracting the different Objective-C implementations (clang does because I wrote them, before Apple added support for their runtime). Instead, when you build Swift, you have a bunch of #ifdef __APPLE__ things that turn off the Objective-C interop for non-Apple platforms. This also means some quite big ABI changes, for example the Swift CoreFoundation implementation has different object layouts to the ones that Apple uses and the way that casts are handled in Swift is different if Swift objects are allowed to secretly be Objective-C objects.
Well, at least they stopped desecrating Objective-C with Java syntax.
Since Steve Naroff left, I don't think anyone on Apple's toolchain team has really understood what made Objective-C good: a simple and clear object model, orthogonal syntax, and good C/C++ interop.
IMHO, although it's sad they completely removed the dynamic aspect of ObjC, swift is the best of all the toy languages. I was comfortable using it to write a full program right away and I do mostly embedded C. The only downside is that it is a bit of an NP-hard problem to decide whether a correct program will compile after adding a line break.
The hard problems in designing a modern programming language are concurrency, security, and error handling. Swift didn't even really try to address any of these (they've now added a few approaches to error handling, but it needed to be baked into the language and consistently applied throughout the standard library from day one). As such, I don't really see it as a language for the 21st century. It gives me similar abstractions to C++ (which, since C++11, has gone from being a horrible language to being a tolerable one), yet with less widespread support and worse performance.
You might want to read up on the actual history a bit more. Monopoly is based on a came called The Landlord's Game, which has two modes of play. In one, you won by constructing a monopoly, in the other you won by increasing the total size of the economy. The point was to illustrate how unconstrained capitalism would lead to monopolies and negative outcomes for most participants.
The modern version is a set of incremental changes to the 1933 game by Parker (later bought by Hasbro). This was a simplified version of The Landlord's Game, which eliminated the cooperative mode and left outright competition as the only objective.
GNUstep has an implementation of the core GUI frameworks (AppKit and a few others), but the open source version of Swift doesn't have Objective-C interop, which is really the only compelling feature for Swift: it's a good incremental language if you have a lot of legacy Objective-C code.
The thought is that embedded wifi will only ever be a security risk, and could never possibly be of use to anyone.
No, the thought is that embedded WiFi is a security risk and this risk outweighs the benefits. A typical fridge lasts, what, 10-20 years? Do you think LG is going to be back-porting network stack security fixes to Linux for 20 years? Do you think that, even if they wanted to, they will make enough profit on fridges to be able to afford to? Over the last 3-4 years, I've lost track of the number of vulnerabilities that enable anyone who can send a packet to the stack to gain kernel-level privilege. Will LG be fixing all of these for the lifetime of the fridge?
Crowding around it and buying it are two different things. I just bought a new fridge and the most important aspects were size, noise, power consumption (running costs) and price. If I'd seen one with an LCD in a shop, I'd have gone and poked it for novelty value, but it wouldn't have contributed to my purchasing decision (except if it came with any kind of network connectivity to check that it came with a 20 year support contract for security updates).
Steve Jobs' Apple presentations were always a little bit too slick for my taste. You can find him introducing various NeXT technologies on YouTube though, and those are well worth watching. The same style that you'll recognise from the Apple presentations, just not yet quite as polished. Also, not presenting mainstream consumer products, so much more interesting (of very dated) tech.
It did have a big capacitive touchscreen, but it wasn't even the first phone to do that (though it was a very close-run thing). The iPhone didn't really become interesting until the second version, when they had third-party apps and 3G. The first version was a high-end feature phone that appealed to people who had never owned a smartphone.
So, in this CS degree, did you learn any actual computer science, or did you just pick up specific technologies that will be obsolete in a decade? From that list of things, it sounds like you got a software engineering qualification from a trade school, not a computer science degree.
The important question to ask is how much will it harm the third party if you go out of business. For pretty much anyone using Google services, the answer is 'not at all'.
You do need a microphone, but most people carry a mobile phone and don't look too carefully at the list of apps that are allowed access to the microphone, GPS and WiFi location data, and the network. If you are a state-level adversary, then compromising, say, the Facebook app will let you get this information in a lot of places.
Apple thought people would want that behaviour because that's the behaviour from MacOS 1-9 and people complained a lot when they broke it in 10.0 (and several times later). Now that the die-hard Mac Classic fans are gone, there's less of a reason to care.
That's because the icon view in the Finder is a spacial view. There is no implicit ordering (and it will remember where the icons are in the window), so selecting two files and expecting it to select a range doesn't make sense. I'm slightly surprised that anyone uses a Finder mode other than the NeXT-like browser mode - the spacial view is for people coming from MacOS 9 and hasn't scaled well to modern filesystem sizes for over a decade.
I agree that this sucks, but the fault is with Intel, not Apple. The only quad-core mobile chips that Intel is currently shipping support a maximum of 16GB if you use DDR3 or LPDDR3, or 32GB if you use DDR4 (and don't support LPDDR4). The difference in power between 16GB of LPDDR3 and 32GB of DDR4 is huge and would take 2-3 hours from the battery life (and have a big impact even when in suspend mode, because the RAM remains powered unless you suspend to disk and pay a wake-up time penalty). The next revision ought to support 32GB of LPDDR4, which should be a nice improvement.
The first new PC I bought was an 8086. Over the years parts got upgraded, including motherboards on occasion. A motherboard is a component upgrade, not a whole new PC, so my statement stands
My first PC was an 8086, which I upgraded with a NEC V30H, but no part of that system other than the ISA bus was standard and so it was basically impossible to upgrade beyond sticking a bit more RAM in. Motherboard form factors weren't standardised until the 486, so even keeping the case was usually impossible, and most 486s didn't come with sufficiently powerful power supplies to be upgraded to pentium.
A motherboard is a component upgrade, not a whole new PC, so my statement stands.
It is effectively a whole new PC is the new motherboard requires a new CPU, GPU, and RAM.
Most non-shit laptops have upgradeable CPUs and GPUs
Okay, name one.
I would love a legal requirement that any device that comes with a camera or microphone have a physical switch to disconnect them.
Another variant of this attack used several other mechanisms for generating the sound. If you're doing a very targeted attack, spiking the CPU to 100% until the fans come on and then letting the machine cool gives you a good idea who it is. For a lot of machines, various different operation sequences can make some components emit high-frequency sound that a reasonable microphone can pick up. There was a really neat attack on Tor in data centres about a decade ago that monitored the ambient temperature (using a co-located box's own temperature monitors) to correlate Tor traffic from a particular node with warming of the room, so that (after a lot of samples) you could tell if traffic for a particular user was flowing through the data centre that you were looking at.
That's funny, because when I used to use a desktop PC, I upgraded it piece by piece, never a full PC at once. Worked fine for decades
I find the 'for decades' part of this hard to believe. The first new PC I bought was a 166MHz Pentium clone. Upgrading the CPU required a new motherboard, because newer chips ran at a different voltage. The new motherboard needed PC100 RAM, not PC66, so I also needed to upgrade the RAM. I kept the case, graphics card, hard disk, CD drive, and floppy, but it was effectively a new computer. The next upgrade required a new case (because you couldn't buy AT form-factor motherboards anymore) for the new motherboard. Again, the new CPU needed new RAM, so I was keeping the drives and GPU, but basically nothing else - I also had to replace my ISA NIC with a PCI one, because there were no ISA slots on the new motherboard. The motherboard upgrade after that kept the case, but the drives needed replacing (no IDE on the new motherboard, only SATA), as did the GPU (no AGP, only PCIe).
The times when I did keep something from the old machine, aside from the case that thing quickly ended up being the bottleneck and was replaced shortly after. Eventually, I realised it was cheaper and less effort to just buy a new machine and demote the old one to some other purpose (often meaning giving it to someone else). The only reason for building my own machine now is when I want something that is too niche for any vendor to sell (e.g. the NAS / media centre box that I have running FreeBSD). For a mobile phone, the niche thing that I want is a minimum of 3-5 years of timely security updates for an Android phone and being able to build the hardware myself wouldn't really help there.
Hell, my current laptop has upgraded RAM modules, disk drive, CPU and GPU.
What laptop has an upgradeable CPU and GPU? The only ones I've seen that don't have both soldered onto the board qualify more as luggables than laptops.