IBM has something called the Scrollpoint mouse, which also gives you 4-way scrolling. There are several other mice with 4-way scrolling as well (tiltable scroll wheel, etc.). Those came out around the same time as the scroll wheel.
Before that, there were several three button mice with the middle button rebound to scrolling in software (you still get that function today in Windows if you like). I think that's actually perhaps the nicest way of doing scrolling.
As for MS porting their NT kernel/APIs to the PDA that is actually a good thing. Who wants to learn a different set of APIs and tools when they don't have to?
But MS did not port the NT kernel or APIs to the PDA. Instead, they created a half-hearted derivative that is a pain to program even for Win32 programmers. That is in addition to the fact that NT was never designed for embedded systems.
Linux, on the other hand, has the full desktop APIs on PDAs. Furthermore, its POSIX APIs have been used in embedded systems for decades.
The upshot is that Linux and POSIX scale well, from PDAs to supercomputers, while NT does not scale well, either up or down.
MS was ahead of the curve in realizing these are the things we would want to do.
MS wasn't ahead of any curve: they chose to deliver a stripped down kernel and set of APIs for PDAs. Even if they had delivered a full handheld system, they would have been far from the first to do so.
WinCe/mobile has PalmOS beat on one thing only; natural handwriting recognition.
Even that depends on what you actually want from your handwriting engine. The PPC handwriting engine is easy to learn and does fairly well on normal text. But it is far from the fastest handwriting input method around, in particular if you need letter accurate input of things like names.
If you look at Macintosh, BSD, and Linux distributions, they also have regular security updates, with many similar vulnerabilities.
There are really two problems here, one true of all major OSes right now, and the other one true of proprietary systems.
The first problem is the pervasive use of C and C++, which makes systems unnecessarily prone to buffer overflows and related problems. C and C++ programmers keep saying that they can handle it, but it is obvious that they can't.
The second problem is that Microsoft and Apple only update their own applications; users are saddled with downloading updates for other software by hand. If all these bugs exist in IE, you can be similar bugs exist in Photoshop, Office, and many other apps that aren't automatically updated.
Windows XP's slower adoption illustrates Microsoft's difficulty in competing with the popularity of its own software platform,
Actually, it probably more illustrates the fear and loathing companies have of the cost, downtime, and productivity losses involved in any Windows upgrade.
It is more structured, which facilitates viewing in an XML capable web browser and editing in an XML editor.
I have yet to see a usable XML editor. And I see no reason to browser kernel configurations in a web browser.
I think doing kernel configuration in XML could still be good, but only if the XML is designed to be human readable in a text editor. And the purpose would not be to use an XML editor, but to permit better manipulation of kernel configurations by scripts.
I'm sure big iron will be important in the computer industry for many decades to come, but it is not where the innovation happens. With Solaris, Sun is investing money in perfecting a dying technology.
The future is light kernels running on hypervisors and clusters. If anything, Linux and BSD need to lose weight and functionality, not gain it.
Despite what many/.ers think, Windows does work well enough to allow people to do productive work.
Even Windows 3.1 satisfied that criterion. The problems with Windows are, and have always been, the costs and risks of going with a proprietary single-vendor solution. The many security and technical issues Windows has are just one expression of those underlying problems.
The problem with comparisons is that once all of the products begin to operate at a level that makes them useful to their target audience
Quite right.
then the only thing left to argue about is the margins.
Not at all: the size and composition of the target audience become the primary focus of discussion. For example, is the fact that Macintosh market share is at 2-3% a result of evil market manipulations, or does it reflect the fact that Macintosh only satisfies the needs of 2-3% of the computer using population? Is the relative usage of BSD and Linux a result of Linux serving more people's needs, or is it a historical accident? Those are the real issues.
Zealots exist on the margins and so are they are the most likely to carp and moan about the small differences between various products. [...] Linus is not a zealot. He is an advocate.
You left out one important group: critics. Unfortunately, zealots often assume that critics are zealots for other causes, but they aren't. It is well worth thinking about what is wrong with Linux, BSD, Macintosh, Windows, etc., and how to do better in the long term. So, someone should point out the problems in BSD, it just shouldn't be Linus.
One key principle of biometric systems is that the system has to verify that the biometric identifier is measured on a living human being. In many settings, that is done automatically (e.g., when a person uses biometric identifiers to verify themselves in person), in other cases, it requires extra technology.
Mercedes screwed up in deploying a bad biometric system, but that doesn't make all biometric systems bad. Your conclusion should be not to buy any more cars from Mercedes (obviously, their engineering department doesn't quite know what they are doing), not to oppose all uses of biometric identifiers.
The entire Zaurus PDA line, the line of TomTom Navigators (probably the best navigation systems you can buy), a number of high-end MP3 and media players, and half a dozen cell phones. And the next generation of Palm will be based on Linux as well.
I enjoy tinkering with my PC OS but when it comes to my PDA it has to just work and it is for getting serious work done quickly. It is the device I turn to when all else fails to get the job done.
There are dozens of handheld devices with Linux pre-installed and supported (often in innovative form factors and applications). There is nothing to tinker: they turn on and are rock solid. They also have lots of software available for them.
Handhelds.org is an effort to bring Linux to unsupported devices, in addition to the supported devices. Why do people do that? Because they can and because they like to have an even broader range of hardware available to them. Generally, installing Linux on unsupported devices involves some effort, but in my experience, the end result is as solid as it is on supported devices.
Actually, I'm quite sure Apple designs large parts of the motherboard itself. Not that it matters as far as my point is concerned:
Indeed, it doesn't matter, because the drivers aren't written for the motherboard, they are written for the components on it.
Red Hat usually doesn't know what computer you run their system on.
No, but companies like Penguin Computing do.
Simple things like support for the particular wireless card inside those notebooks is a constant problem, and some of my pals (mostly those running Linux) have actually adopted to this by always using ethernet for Internet access, even though they own wifi cards.
With Linux, you have a choice: you can buy supported hardware, in which case the WiFi card will work, or you can make a conscious decision to buy unsupported hardawre. It's your/their choice.
For Mac users, it's a non-issue.
It is very much an issue for Mac users: they have to be happy with the two kinds of laptops and three screen sizes Apple gives them, whereas Linux and Windows users can choose among dozens of different models.
It's not a statistic, it's just my own experience which matches what you logically would expect, given Apple's business model of making "the whole widget".
If you want Apple-like hardware/software integration, you can get it for Linux and Windows from multiple vendors. But Apple's miniscule market share alone shows that that simply isn't a big issue for most buyers.
How difficult do you think it will be to hack the 'smart chip' held on the card ?
I know it will be a lot harder than printing a fake card on a laser printer or hacking the magnetic strip on current cards.
Better a simple human operated system powered by suspicioius old ladies, than a complex technological one operated by fallible machines.
Suspicious old ladies are getting to be rarer and rarer in the service industries. If you want to engage in stereotyping, it's more the 20-something bubble-gum chewing untrained clerks that I'm worried about. But even the suspicious old ladies are demonstrably not very good at detecting fakes and frauds, while correctly implemented biometrics is.
How sensitive do you think the biometric match will be in order to minimise the false negative matches that will hold up the queue in the bank / post office ?
I think it is nice that with biometric technologies, one can actually ask the question and make deliberate tradeoffs, as opposed to leaving it to the whims of some low-paid clerk.
I think we do share you concern that companies and governments may not implement biometrics correctly. But the solution to that problem is to demand transparency and open systems, not to avoid biometrics altogether.
Biometric identifiers aren't keys--they don't have to be "revocable". Unlike a key, knowing the digital form of a biometric identifier does not let you authenticate.
What is somebody going to do with your driver's license?
Unauthorized use of my driver's license lets people open a bank account and ruin my credit rating (driver's license in the US = ID card in the UK), among many other things.
We're blocking New Labour's plans for mass surveillance and data retention, 1984 style.
I wish you were, but you are not. You are, instead, wasting time with something that gets you a lot of press coverage but actually makes the situation worse.
Mass surveillance and data retention are already here, without national ID cards. But if people are going to retain data on my life, at least I would like to make it harder for other people to impersonate me.
Think of your last software project: deadlines, last minute bug fixes, etc. Are API specs for your software ready in releasable form along with your software?
Well, it's no different for a lot of hardware: just because a company got silicon and just because someone managed to hack together a Windows driver that (barely) works doesn't mean they have anything approaching releasable specs that others can use to build drivers.
Sorry, guys, but I want smartcard identification and biometric identifiers. I find it ridiculous that if my driver's license or credit card gets stolen, someone can trivially impersonate me, wreck my credit rating, and do other things to ruin me. Given the widespread availability of high quality scanners and printers, paper and plastic just won't do anymore.
Of course, we shouldn't kid ourselves: smartcards and biometric identifiers are not sufficient for improving security, and they will do little to stop terrorism. But, while not sufficient, they are a necessary component of any future system.
in the limit, the packaging, power, network, and thermal costs will drive the individual nodes in any cluster to be 'fatter'.
Quite to the contrary, once you are willing to use clustering, it makes sense to make the individual nodes a bit leaner: you reduce packaging, power, network, and thermal costs that way.
It doesn't take an MIT scientists to do those things. Go and look at hobbyist magazines on woodworking and metalworking: they are full of these kinds of computer-controlled tools. It's kind of ironic that good old American hobbies are being sold by futurists and scientists as the next great thing.
However, all of those devices are still far from being "desktop fabs": they cannot create complex machinery, they require manual intervention, they require expertise to operate, they require expensive manufactured manufactured materials, and they certainly cannot replicate themselves. It will take a lot of engineering to address those problems, and that kind of engineering will not come from a bunch of publicity-hungry futurists.
If you review the scientific process again you will find that it does not allow for tradeoffs.
There are millions of experiments one could do at any one time. The ones that get done are the ones that are economically feasible and get funded. Those are the tradeoffs working scientists make every day.
[the hardware, software, expansion hardware, and software upgrades ] Maybe I should have been more specific. It's developed by the same entity.
Have you looked in your Mac lately? Macs these days consist almost completely of standard PC hardware (and are, in fact, built by contract PC manufacturers in the far east). And it is very unlikely that Apple develops all its drivers from scratch.
Apple selects from standard components, contracts out assembly, and then adapts their software platform to run on those machines, reusing as much driver code from vendors as they can. That's the same thing a good PC manufacturer that ships Linux or Windows machines does.
Sorry, there just is no difference in the way Apple operates. If you want to claim that Apple does a better job, maybe you have some statistics to back that up? My own experience with Apple products has been generally to other name-brand PC manufacturers, no better and no worse.
I agree that multithreading is not the future, for all the reasons you give. But I don't believe multicell is either: you still have memory and I/O bottlenecks.
In fact, the future of high performance computing is already here: large amounts of commodity hardware. Every box you add automatically adds not only another CPU, but also a separate memory system and I/O.
Having said that, multicell doesn't have a future as a general purpose parallel computing paradigm, but it does hold the promise of being able to replace GPUs and other special-purpose hardware that litters our machines right now.
Not trying to camp fanbois here but if you really need "down to the metal" performance you're writing in C with assembler hotspots.
And you are going to hand-tune your assembly language hotspots and your C code to work on every single variant of the x86 architecture? I think not. Java isn't the answer to high performance computing, but neither are C or assembly. Some kind of JIT is likely the future, together with a high-level language that sucks less than Java.
Slashdot Mirror
So, it sounds like this is both a new format and a new compressor.
Well, the immediate questions are:
* How good is the compressor? Say, relative to Xvid, for example? Is it still fully MPEG4 compatible?
* Is the DMF format open, closed, or even patented?
IBM has something called the Scrollpoint mouse, which also gives you 4-way scrolling. There are several other mice with 4-way scrolling as well (tiltable scroll wheel, etc.). Those came out around the same time as the scroll wheel.
Before that, there were several three button mice with the middle button rebound to scrolling in software (you still get that function today in Windows if you like). I think that's actually perhaps the nicest way of doing scrolling.
As for MS porting their NT kernel/APIs to the PDA that is actually a good thing. Who wants to learn a different set of APIs and tools when they don't have to?
But MS did not port the NT kernel or APIs to the PDA. Instead, they created a half-hearted derivative that is a pain to program even for Win32 programmers. That is in addition to the fact that NT was never designed for embedded systems.
Linux, on the other hand, has the full desktop APIs on PDAs. Furthermore, its POSIX APIs have been used in embedded systems for decades.
The upshot is that Linux and POSIX scale well, from PDAs to supercomputers, while NT does not scale well, either up or down.
MS was ahead of the curve in realizing these are the things we would want to do.
MS wasn't ahead of any curve: they chose to deliver a stripped down kernel and set of APIs for PDAs. Even if they had delivered a full handheld system, they would have been far from the first to do so.
WinCe/mobile has PalmOS beat on one thing only; natural handwriting recognition.
Even that depends on what you actually want from your handwriting engine. The PPC handwriting engine is easy to learn and does fairly well on normal text. But it is far from the fastest handwriting input method around, in particular if you need letter accurate input of things like names.
If you look at Macintosh, BSD, and Linux distributions, they also have regular security updates, with many similar vulnerabilities.
There are really two problems here, one true of all major OSes right now, and the other one true of proprietary systems.
The first problem is the pervasive use of C and C++, which makes systems unnecessarily prone to buffer overflows and related problems. C and C++ programmers keep saying that they can handle it, but it is obvious that they can't.
The second problem is that Microsoft and Apple only update their own applications; users are saddled with downloading updates for other software by hand. If all these bugs exist in IE, you can be similar bugs exist in Photoshop, Office, and many other apps that aren't automatically updated.
Windows XP's slower adoption illustrates Microsoft's difficulty in competing with the popularity of its own software platform,
Actually, it probably more illustrates the fear and loathing companies have of the cost, downtime, and productivity losses involved in any Windows upgrade.
It is more structured, which facilitates viewing in an XML capable web browser and editing in an XML editor.
I have yet to see a usable XML editor. And I see no reason to browser kernel configurations in a web browser.
I think doing kernel configuration in XML could still be good, but only if the XML is designed to be human readable in a text editor. And the purpose would not be to use an XML editor, but to permit better manipulation of kernel configurations by scripts.
I'm sure big iron will be important in the computer industry for many decades to come, but it is not where the innovation happens. With Solaris, Sun is investing money in perfecting a dying technology.
The future is light kernels running on hypervisors and clusters. If anything, Linux and BSD need to lose weight and functionality, not gain it.
I don't think even the most hard-core Linux user would dispute that (well, maybe the zealots would).
A Ferrari may be the "best" car there is according to some criteria, but that doesn't make it a good car for most people.
Despite what many /.ers think, Windows does work well enough to allow people to do productive work.
Even Windows 3.1 satisfied that criterion. The problems with Windows are, and have always been, the costs and risks of going with a proprietary single-vendor solution. The many security and technical issues Windows has are just one expression of those underlying problems.
The problem with comparisons is that once all of the products begin to operate at a level that makes them useful to their target audience
Quite right.
then the only thing left to argue about is the margins.
Not at all: the size and composition of the target audience become the primary focus of discussion. For example, is the fact that Macintosh market share is at 2-3% a result of evil market manipulations, or does it reflect the fact that Macintosh only satisfies the needs of 2-3% of the computer using population? Is the relative usage of BSD and Linux a result of Linux serving more people's needs, or is it a historical accident? Those are the real issues.
Zealots exist on the margins and so are they are the most likely to carp and moan about the small differences between various products. [...] Linus is not a zealot. He is an advocate.
You left out one important group: critics. Unfortunately, zealots often assume that critics are zealots for other causes, but they aren't. It is well worth thinking about what is wrong with Linux, BSD, Macintosh, Windows, etc., and how to do better in the long term. So, someone should point out the problems in BSD, it just shouldn't be Linus.
One key principle of biometric systems is that the system has to verify that the biometric identifier is measured on a living human being. In many settings, that is done automatically (e.g., when a person uses biometric identifiers to verify themselves in person), in other cases, it requires extra technology.
Mercedes screwed up in deploying a bad biometric system, but that doesn't make all biometric systems bad. Your conclusion should be not to buy any more cars from Mercedes (obviously, their engineering department doesn't quite know what they are doing), not to oppose all uses of biometric identifiers.
The entire Zaurus PDA line, the line of TomTom Navigators (probably the best navigation systems you can buy), a number of high-end MP3 and media players, and half a dozen cell phones. And the next generation of Palm will be based on Linux as well.
I enjoy tinkering with my PC OS but when it comes to my PDA it has to just work and it is for getting serious work done quickly. It is the device I turn to when all else fails to get the job done.
There are dozens of handheld devices with Linux pre-installed and supported (often in innovative form factors and applications). There is nothing to tinker: they turn on and are rock solid. They also have lots of software available for them.
Handhelds.org is an effort to bring Linux to unsupported devices, in addition to the supported devices. Why do people do that? Because they can and because they like to have an even broader range of hardware available to them. Generally, installing Linux on unsupported devices involves some effort, but in my experience, the end result is as solid as it is on supported devices.
Actually, I'm quite sure Apple designs large parts of the motherboard itself. Not that it matters as far as my point is concerned:
Indeed, it doesn't matter, because the drivers aren't written for the motherboard, they are written for the components on it.
Red Hat usually doesn't know what computer you run their system on.
No, but companies like Penguin Computing do.
Simple things like support for the particular wireless card inside those notebooks is a constant problem, and some of my pals (mostly those running Linux) have actually adopted to this by always using ethernet for Internet access, even though they own wifi cards.
With Linux, you have a choice: you can buy supported hardware, in which case the WiFi card will work, or you can make a conscious decision to buy unsupported hardawre. It's your/their choice.
For Mac users, it's a non-issue.
It is very much an issue for Mac users: they have to be happy with the two kinds of laptops and three screen sizes Apple gives them, whereas Linux and Windows users can choose among dozens of different models.
It's not a statistic, it's just my own experience which matches what you logically would expect, given Apple's business model of making "the whole widget".
If you want Apple-like hardware/software integration, you can get it for Linux and Windows from multiple vendors. But Apple's miniscule market share alone shows that that simply isn't a big issue for most buyers.
How difficult do you think it will be to hack the 'smart chip' held on the card ?
I know it will be a lot harder than printing a fake card on a laser printer or hacking the magnetic strip on current cards.
Better a simple human operated system powered by suspicioius old ladies, than a complex technological one operated by fallible machines.
Suspicious old ladies are getting to be rarer and rarer in the service industries. If you want to engage in stereotyping, it's more the 20-something bubble-gum chewing untrained clerks that I'm worried about. But even the suspicious old ladies are demonstrably not very good at detecting fakes and frauds, while correctly implemented biometrics is.
How sensitive do you think the biometric match will be in order to minimise the false negative matches that will hold up the queue in the bank / post office ?
I think it is nice that with biometric technologies, one can actually ask the question and make deliberate tradeoffs, as opposed to leaving it to the whims of some low-paid clerk.
I think we do share you concern that companies and governments may not implement biometrics correctly. But the solution to that problem is to demand transparency and open systems, not to avoid biometrics altogether.
Biometric identifiers aren't keys--they don't have to be "revocable". Unlike a key, knowing the digital form of a biometric identifier does not let you authenticate.
What is somebody going to do with your driver's license?
Unauthorized use of my driver's license lets people open a bank account and ruin my credit rating (driver's license in the US = ID card in the UK), among many other things.
We're blocking New Labour's plans for mass surveillance and data retention, 1984 style.
I wish you were, but you are not. You are, instead, wasting time with something that gets you a lot of press coverage but actually makes the situation worse.
Mass surveillance and data retention are already here, without national ID cards. But if people are going to retain data on my life, at least I would like to make it harder for other people to impersonate me.
Think of your last software project: deadlines, last minute bug fixes, etc. Are API specs for your software ready in releasable form along with your software?
Well, it's no different for a lot of hardware: just because a company got silicon and just because someone managed to hack together a Windows driver that (barely) works doesn't mean they have anything approaching releasable specs that others can use to build drivers.
Sorry, guys, but I want smartcard identification and biometric identifiers. I find it ridiculous that if my driver's license or credit card gets stolen, someone can trivially impersonate me, wreck my credit rating, and do other things to ruin me. Given the widespread availability of high quality scanners and printers, paper and plastic just won't do anymore.
Of course, we shouldn't kid ourselves: smartcards and biometric identifiers are not sufficient for improving security, and they will do little to stop terrorism. But, while not sufficient, they are a necessary component of any future system.
in the limit, the packaging, power, network, and thermal costs will drive the individual nodes in any cluster to be 'fatter'.
Quite to the contrary, once you are willing to use clustering, it makes sense to make the individual nodes a bit leaner: you reduce packaging, power, network, and thermal costs that way.
It doesn't take an MIT scientists to do those things. Go and look at hobbyist magazines on woodworking and metalworking: they are full of these kinds of computer-controlled tools. It's kind of ironic that good old American hobbies are being sold by futurists and scientists as the next great thing.
However, all of those devices are still far from being "desktop fabs": they cannot create complex machinery, they require manual intervention, they require expertise to operate, they require expensive manufactured manufactured materials, and they certainly cannot replicate themselves. It will take a lot of engineering to address those problems, and that kind of engineering will not come from a bunch of publicity-hungry futurists.
If you review the scientific process again you will find that it does not allow for tradeoffs.
There are millions of experiments one could do at any one time. The ones that get done are the ones that are economically feasible and get funded. Those are the tradeoffs working scientists make every day.
[the hardware, software, expansion hardware, and software upgrades ] Maybe I should have been more specific. It's developed by the same entity.
Have you looked in your Mac lately? Macs these days consist almost completely of standard PC hardware (and are, in fact, built by contract PC manufacturers in the far east). And it is very unlikely that Apple develops all its drivers from scratch.
Apple selects from standard components, contracts out assembly, and then adapts their software platform to run on those machines, reusing as much driver code from vendors as they can. That's the same thing a good PC manufacturer that ships Linux or Windows machines does.
Sorry, there just is no difference in the way Apple operates. If you want to claim that Apple does a better job, maybe you have some statistics to back that up? My own experience with Apple products has been generally to other name-brand PC manufacturers, no better and no worse.
Multi-threading is NOT the future. Multi-cell is.
I agree that multithreading is not the future, for all the reasons you give. But I don't believe multicell is either: you still have memory and I/O bottlenecks.
In fact, the future of high performance computing is already here: large amounts of commodity hardware. Every box you add automatically adds not only another CPU, but also a separate memory system and I/O.
Having said that, multicell doesn't have a future as a general purpose parallel computing paradigm, but it does hold the promise of being able to replace GPUs and other special-purpose hardware that litters our machines right now.
Not trying to camp fanbois here but if you really need "down to the metal" performance you're writing in C with assembler hotspots.
And you are going to hand-tune your assembly language hotspots and your C code to work on every single variant of the x86 architecture? I think not. Java isn't the answer to high performance computing, but neither are C or assembly. Some kind of JIT is likely the future, together with a high-level language that sucks less than Java.