Microsoft was the first ever company to produce an OS, or where there companies before that went bust despite at one point being market leader?
Microsoft was the first (I think, certainly one of the first) company to sell an OS as a commodity off-the-shelf product. Previously, operating systems had been sold to computer manufacturers, tailored specifically to their system, and developed in-house with the computer. IBM was the first company to develop an OS (OS/360) that ran on multiple computers and allowed userspace software to be trivially ported between them. Microsoft recognised this trend (almost 20 years later) and realised that the OS, not the hardware, was what defined the computer and that this was the right monopoly to aim for.
As for his story "All you Zombies", don't forget that not only did the narrator do his mother, he was his mother!
I was actually thinking of Number of the Beast and Time Enough for Love (which could have been more accurately titled Time Enough for Eugenics). I forgot about All you Zombies, although at least in that case it's not the character's goal and motivation.
If it were something capable of crossing interstellar space, then the thruster exhaust (even if they had some exotic drive for travelling very long distances) would likely be invisible in the visible spectrum and very, very bright in the IR. It would also likely be travelling significantly faster than anything naturally occurring and so would be quite distinctive in several ways.
The criteria could be — from Heinlein's other writings — an ability to solve a linear (or square) equation, for example
Look up how Jim Crow laws worked in the US. Even literacy tests were used for racial discrimination, and it's easy for that kind of thing to become self-perpetuating. If the poor people only have access to public schooling and it isn't very good, then you can write the tests so that only people who had a better education can pass it and then defund public schools more...
I'd be in favour of a simpler test: Do you know what the candidate stands for? Can you pass a simple multiple choice tests about the candidates policies, prepared by the candidate and checked against the promises that they made during their campaign by the electoral committee? The candidate would be free to hand out solutions to the test on their campaign leaflets, as long as the voters actually read them and learn more than 'he wants to lower taxes / is {for, against} abortion'.
This is an interesting change from the book, because the scene is almost exactly the same but the meaning is totally changed (once you get another chapter in they diverge to the extent that it's impossible to tell they they're even similar stories). In the book, he's in the recruiting office to discourage people from signing up with any rosy view of what they're getting in to. When he leaves, he puts on prosthetics that make him seem completely normal - the mutilated veteran appearance is just for show.
There's a good reason why the film diverged from the book - the book just isn't that good. The film is a satire of what Heinlein wrote in total seriousness. His books are a mixture of cult-of-the-individual libertarianism and characters travelling back in time so that they can fuck their mother[1]. It must be incredibly hard to write a screenplay based on his work that isn't satire, because there's no way you can take it seriously.
[1] Yes, he really did write two books about this.
I don't think we can blame DICE for this. SourceForge has been increasingly bad for years. The real question is why DICE thought SourceForge was worth anything. Ten years ago, looking for open source stuff (or looking for somewhere to host a new project), SourceForge would have been near the top of my list. Now, and for the last few years, whenever I see a SourceForge link I expect to see either an abandoned project or a site that's totally impossible to navigate to find what I want.
The difference between US and English culture can be seen quite clearly in the difference between baseball and cricket. Baseball has regular breaks for adverts, cricket has regular breaks for tea.
Your argument seems to be that you can't have patents on simple things, but software can be much more complex than mechanical systems that can be patented. Dyson, for example, has a patent on a particular curvature of pipe that allows air to flow and prevents things being carried along with it from being snagged. This is pretty simple, and it's also basically mathematics (it's the solution to a particular computational fluid dynamics model). In contrast, you'd say that something like a wavelet-based compression algorithm should not be patented, because it's 'just math', even though it's orders of magnitude more complex and likely not obvious to someone skilled in the art.
You don't have to pay the patent license to have the filesystem, only for software that can run it. You also don't need to install software unless you want the device to appear as a native filesystem. You can easily just use the USB generic device interface and provide userspace software for communicating. Better still, why not make it appear as a USB network interface and run an SMB server or similar?
There are lots of good arguments against software patents, but this one is just bullshit. Software can be represented in mathematical terms and programming languages are can be considered mathematical languages, but the same is true of blueprints and pretty much any other form of design. If 'it's math' is an argument against software patents then it's also an argument against all other patents.
Taxis can be cheaper than driving, depending on the amount of driving you need to do. I make a trip where a car would be useful maybe once a month (and a lot of those are ending at an airport, where if I drove myself I'd have huge parking costs as well). I have a 10-minute bike ride to work each day, and on the same roads I'd be slower to drive (lots of traffic lights and a cycle path that doesn't have them, and huge traffic jams at peak times that make a car only slightly faster than walking).
The fixed costs of owning a car are quite high, but the per-mile cost on top of that is significantly lower than a taxi. If you drive enough that the incremental costs are the dominant factor, then a car is cheaper. If you drive sufficiently infrequently that the fixed costs (insurance, initial purchase / depreciation, tax) are dominant then a taxi can be cheaper. In my case, it would cost more to just own a car, even if I never drove it, than to use taxis and trains.
Would you... feel better if we privatized the NSA and called it Facebook 2.0? Heh. Hehehe. hehehehehehheeeeee.... idiots.
Well, the CIA would be disappointed. The fun thing about the Google vs Facebook competition is the undertones of interservice rivalry. On the one hand, you have Google that employs a lot of ex-NSA people (and the NSA, that employs a lot of ex-Google people - for a while, they were the only two big employers of data mining specialists in the USA) and on the other hand you have Facebook that got a lot of its initial funding from the CIA. I bet the CIA is now really upset to learn that the NSA has had better access to Facebook data than them...
Depends on what you mean by 'cleaned up'. x86-64 replaced the old RISC servers, but they died because the companies producing them stopped developing them. And that happened because they'd all bet on Itanium.
Compaq/HP should have left NonStop on MIPS and VAX on Alpha, and never gone the Itanic route w/ them
You're missing HP-UX on PA-RISC and Tru64 on Alpha. This is why HP jumped on the Itanium bandwagon. They weren't big enough to be viable developing one CPU for in-house use and they had two (PA-RISC and Alpha) that they needed for their own product lines. Killing one and migrating to the other would not have gone down well with their customers or engineers, and wouldn't have solved the problem that successful microprocessor development is all about volumes: it doesn't matter if you've got a chip for a niche architecture that's twice as fast as Intel's, because they'll sell well over a hundred times as many as you and so you won't be able to compete on performance per dollar. IBM only manages it by sharing a huge amount of their designs between mainframe, supercomputer, and embedded chips, and even then the POWER line struggles relative to the Xeon - it's competitive, but it's not a clear winner.
DEC paid Microsoft to port Windows to Alpha. If they'd also paid them to port Office and Visual Basic, and had given away free Alpha workstations to the top thousand Windows software development companies, then we'd probably all be using Alphas now. They missed the boat, and HP had to deal with the situation as they found it.
The problem was not that they decided to go with an outsourced architecture, it was that the one they picked was a clusterfuck from the start. The philosophy behind the CISC movement was to produce chips that were easy for assembly writers to program. The philosophy behind the RISC movement was to produce chips that were easy for compilers to target. The philosophy behind the VLIW movement was to produce chips that are easy to design, but hard to program for humans and compilers. It failed the first three times Intel tried it, yet they managed to convince HP that this time it would work really well.
A better strategy for HP might have been to follow Apple's example with the Apple-IBM-Motorola group, and set up a new consortium to make workstation and server CPUs. Sun and Fujitsu were already starting to run collaborate on designs. Between them and HP, they might have been able to produce one line of CPUs that would be easy to emulate SPARC, Alpha, and PA-RISC on and which could have kept the performance lead over Intel. By sharing the development costs between the three of them (maybe four, if Motorola had been interested), they'd have been able to push the unit costs down quite a bit.
OS/2 used different rings for kernel, drivers, and userspace.
Novell Netware (pretty much dead now, but very popular in the 386-486 era) allowed plugins to run in rings 1-2 for extra speed without being able to compromise the kernel.
And, the big one, Xen in a PV environment runs in ring 0, PV guests (including dom0) run in ring 1, userspace code runs in ring 3. Ring 2 is usually unused, although it could be used by some guests.
And if we're not counting just the original 4 rings, there are a lot of things that run in the ring -1 that the virtualisation extensions provide.
The funny thing about the whole story is that the reason i386 has four protection rings was to make a VMS port from VAX easier. Instead, DEC ported it to Alpha, with only two...
Even if a process were a complete sandbox, this kind of attack would barely be mitigated, because an exploit in a library allows running arbitrary code (you might want to look up return oriented programming, if you think avoiding code generation helps you). At this point, the person who has sent you an email with a.tiff attachment now has complete control of your mail client.
but the fact that Windows has images and fonts that can own your system is beyond absurd.
It is absurd, but let's not pick on Windows. Both OS X and *NIX systems have suffered from similar vulnerabilities in libtiff, libpng (lots!), libjpeg (almost as many) and FreeType (too many to count). The problem was that all of these were written with the assumption that you could trust the input data and that performance was the primary concern. Now, computers are so fast that no one would notice a 50% slowdown in most of these (although they would in an H.264 decoder, which is another popular vector), and people attack them with fuzzing tools to try to find exploits so the input can't be trusted at all.
Right, because libtiff (and libpng and libjpeg) have never had security issues on Linux that allow a maliciously crafted image to execute arbitrary code. (Hint for those that don't get sarcasm: search the CVE database for any of those and filter by arbitrary code execution vulnerability)
Code signing is far from a panacea. It only works well in a world where there is a clear divide between things that are programs and things that are data. It doesn't help if you sign your interpreter (for Python, VBA, JavaScript, whatever), if there's no requirement that you also sign all of the inputs.
And code signing would do nothing to prevent vulnerabilities of this nature, where a bug in a library permits arbitrary code execution. This can be prevented with fine-grained sandboxing (if every TIFF image were decoded in a totally unprivileged sandbox, this exploit would be mitigated), but that's not feasible with current CPUs.
The burning of Guys atop bonfires is a celebration that the plot failed. The setting off lots of explosives is a celebration of the fact that it could have succeeded and a reminder to our elected officials not to be too complacent. Most of us just watch the fireworks...
Slashdot Mirror
Microsoft was the first ever company to produce an OS, or where there companies before that went bust despite at one point being market leader?
Microsoft was the first (I think, certainly one of the first) company to sell an OS as a commodity off-the-shelf product. Previously, operating systems had been sold to computer manufacturers, tailored specifically to their system, and developed in-house with the computer. IBM was the first company to develop an OS (OS/360) that ran on multiple computers and allowed userspace software to be trivially ported between them. Microsoft recognised this trend (almost 20 years later) and realised that the OS, not the hardware, was what defined the computer and that this was the right monopoly to aim for.
As for his story "All you Zombies", don't forget that not only did the narrator do his mother, he was his mother!
I was actually thinking of Number of the Beast and Time Enough for Love (which could have been more accurately titled Time Enough for Eugenics). I forgot about All you Zombies, although at least in that case it's not the character's goal and motivation.
If it were something capable of crossing interstellar space, then the thruster exhaust (even if they had some exotic drive for travelling very long distances) would likely be invisible in the visible spectrum and very, very bright in the IR. It would also likely be travelling significantly faster than anything naturally occurring and so would be quite distinctive in several ways.
The criteria could be — from Heinlein's other writings — an ability to solve a linear (or square) equation, for example
Look up how Jim Crow laws worked in the US. Even literacy tests were used for racial discrimination, and it's easy for that kind of thing to become self-perpetuating. If the poor people only have access to public schooling and it isn't very good, then you can write the tests so that only people who had a better education can pass it and then defund public schools more...
I'd be in favour of a simpler test: Do you know what the candidate stands for? Can you pass a simple multiple choice tests about the candidates policies, prepared by the candidate and checked against the promises that they made during their campaign by the electoral committee? The candidate would be free to hand out solutions to the test on their campaign leaflets, as long as the voters actually read them and learn more than 'he wants to lower taxes / is {for, against} abortion'.
This is an interesting change from the book, because the scene is almost exactly the same but the meaning is totally changed (once you get another chapter in they diverge to the extent that it's impossible to tell they they're even similar stories). In the book, he's in the recruiting office to discourage people from signing up with any rosy view of what they're getting in to. When he leaves, he puts on prosthetics that make him seem completely normal - the mutilated veteran appearance is just for show.
There's a good reason why the film diverged from the book - the book just isn't that good. The film is a satire of what Heinlein wrote in total seriousness. His books are a mixture of cult-of-the-individual libertarianism and characters travelling back in time so that they can fuck their mother[1]. It must be incredibly hard to write a screenplay based on his work that isn't satire, because there's no way you can take it seriously.
[1] Yes, he really did write two books about this.
I don't think we can blame DICE for this. SourceForge has been increasingly bad for years. The real question is why DICE thought SourceForge was worth anything. Ten years ago, looking for open source stuff (or looking for somewhere to host a new project), SourceForge would have been near the top of my list. Now, and for the last few years, whenever I see a SourceForge link I expect to see either an abandoned project or a site that's totally impossible to navigate to find what I want.
The difference between US and English culture can be seen quite clearly in the difference between baseball and cricket. Baseball has regular breaks for adverts, cricket has regular breaks for tea.
City or United?
If not SMB, then WebDAV - Windows (since Windows 98) can mount WebDAV folders as if they were local.
Your argument seems to be that you can't have patents on simple things, but software can be much more complex than mechanical systems that can be patented. Dyson, for example, has a patent on a particular curvature of pipe that allows air to flow and prevents things being carried along with it from being snagged. This is pretty simple, and it's also basically mathematics (it's the solution to a particular computational fluid dynamics model). In contrast, you'd say that something like a wavelet-based compression algorithm should not be patented, because it's 'just math', even though it's orders of magnitude more complex and likely not obvious to someone skilled in the art.
You don't have to pay the patent license to have the filesystem, only for software that can run it. You also don't need to install software unless you want the device to appear as a native filesystem. You can easily just use the USB generic device interface and provide userspace software for communicating. Better still, why not make it appear as a USB network interface and run an SMB server or similar?
There are lots of good arguments against software patents, but this one is just bullshit. Software can be represented in mathematical terms and programming languages are can be considered mathematical languages, but the same is true of blueprints and pretty much any other form of design. If 'it's math' is an argument against software patents then it's also an argument against all other patents.
exFAT more than FAT, but they also license a lot of patents covering stuff from Wince.
Taxis can be cheaper than driving, depending on the amount of driving you need to do. I make a trip where a car would be useful maybe once a month (and a lot of those are ending at an airport, where if I drove myself I'd have huge parking costs as well). I have a 10-minute bike ride to work each day, and on the same roads I'd be slower to drive (lots of traffic lights and a cycle path that doesn't have them, and huge traffic jams at peak times that make a car only slightly faster than walking). The fixed costs of owning a car are quite high, but the per-mile cost on top of that is significantly lower than a taxi. If you drive enough that the incremental costs are the dominant factor, then a car is cheaper. If you drive sufficiently infrequently that the fixed costs (insurance, initial purchase / depreciation, tax) are dominant then a taxi can be cheaper. In my case, it would cost more to just own a car, even if I never drove it, than to use taxis and trains.
Would you... feel better if we privatized the NSA and called it Facebook 2.0? Heh. Hehehe. hehehehehehheeeeee.... idiots.
Well, the CIA would be disappointed. The fun thing about the Google vs Facebook competition is the undertones of interservice rivalry. On the one hand, you have Google that employs a lot of ex-NSA people (and the NSA, that employs a lot of ex-Google people - for a while, they were the only two big employers of data mining specialists in the USA) and on the other hand you have Facebook that got a lot of its initial funding from the CIA. I bet the CIA is now really upset to learn that the NSA has had better access to Facebook data than them...
Depends on what you mean by 'cleaned up'. x86-64 replaced the old RISC servers, but they died because the companies producing them stopped developing them. And that happened because they'd all bet on Itanium.
Sounds a lot like the Pentium Pro - great 32-bit performance, but terrible 16-bit.
Compaq/HP should have left NonStop on MIPS and VAX on Alpha, and never gone the Itanic route w/ them
You're missing HP-UX on PA-RISC and Tru64 on Alpha. This is why HP jumped on the Itanium bandwagon. They weren't big enough to be viable developing one CPU for in-house use and they had two (PA-RISC and Alpha) that they needed for their own product lines. Killing one and migrating to the other would not have gone down well with their customers or engineers, and wouldn't have solved the problem that successful microprocessor development is all about volumes: it doesn't matter if you've got a chip for a niche architecture that's twice as fast as Intel's, because they'll sell well over a hundred times as many as you and so you won't be able to compete on performance per dollar. IBM only manages it by sharing a huge amount of their designs between mainframe, supercomputer, and embedded chips, and even then the POWER line struggles relative to the Xeon - it's competitive, but it's not a clear winner.
DEC paid Microsoft to port Windows to Alpha. If they'd also paid them to port Office and Visual Basic, and had given away free Alpha workstations to the top thousand Windows software development companies, then we'd probably all be using Alphas now. They missed the boat, and HP had to deal with the situation as they found it.
The problem was not that they decided to go with an outsourced architecture, it was that the one they picked was a clusterfuck from the start. The philosophy behind the CISC movement was to produce chips that were easy for assembly writers to program. The philosophy behind the RISC movement was to produce chips that were easy for compilers to target. The philosophy behind the VLIW movement was to produce chips that are easy to design, but hard to program for humans and compilers. It failed the first three times Intel tried it, yet they managed to convince HP that this time it would work really well.
A better strategy for HP might have been to follow Apple's example with the Apple-IBM-Motorola group, and set up a new consortium to make workstation and server CPUs. Sun and Fujitsu were already starting to run collaborate on designs. Between them and HP, they might have been able to produce one line of CPUs that would be easy to emulate SPARC, Alpha, and PA-RISC on and which could have kept the performance lead over Intel. By sharing the development costs between the three of them (maybe four, if Motorola had been interested), they'd have been able to push the unit costs down quite a bit.
And if we're not counting just the original 4 rings, there are a lot of things that run in the ring -1 that the virtualisation extensions provide.
The funny thing about the whole story is that the reason i386 has four protection rings was to make a VMS port from VAX easier. Instead, DEC ported it to Alpha, with only two...
Even if a process were a complete sandbox, this kind of attack would barely be mitigated, because an exploit in a library allows running arbitrary code (you might want to look up return oriented programming, if you think avoiding code generation helps you). At this point, the person who has sent you an email with a .tiff attachment now has complete control of your mail client.
but the fact that Windows has images and fonts that can own your system is beyond absurd.
It is absurd, but let's not pick on Windows. Both OS X and *NIX systems have suffered from similar vulnerabilities in libtiff, libpng (lots!), libjpeg (almost as many) and FreeType (too many to count). The problem was that all of these were written with the assumption that you could trust the input data and that performance was the primary concern. Now, computers are so fast that no one would notice a 50% slowdown in most of these (although they would in an H.264 decoder, which is another popular vector), and people attack them with fuzzing tools to try to find exploits so the input can't be trusted at all.
Right, because libtiff (and libpng and libjpeg) have never had security issues on Linux that allow a maliciously crafted image to execute arbitrary code. (Hint for those that don't get sarcasm: search the CVE database for any of those and filter by arbitrary code execution vulnerability)
Code signing is far from a panacea. It only works well in a world where there is a clear divide between things that are programs and things that are data. It doesn't help if you sign your interpreter (for Python, VBA, JavaScript, whatever), if there's no requirement that you also sign all of the inputs.
And code signing would do nothing to prevent vulnerabilities of this nature, where a bug in a library permits arbitrary code execution. This can be prevented with fine-grained sandboxing (if every TIFF image were decoded in a totally unprivileged sandbox, this exploit would be mitigated), but that's not feasible with current CPUs.
The burning of Guys atop bonfires is a celebration that the plot failed. The setting off lots of explosives is a celebration of the fact that it could have succeeded and a reminder to our elected officials not to be too complacent. Most of us just watch the fireworks...