It is the nature of a kernel that mods may require patches 'all over the kernel'.
No. It is the nature of big, monolithic programs that don't use modern abstraction facilities to require patches all over. Kernels are no different from word processors in that regard.
That said, your suggestion of using hooks will also require patches 'all over the kernel'.
No. You put in the hooks first, everywhere. Do it from 2.5.10 to 2.5.11. Afterwards, any module can hook just about any function anywhere in the kernel without patches.
Either kernel additions like USB can be distributed without patching the kernel or they can't. If they require patching, it's a burden on Linux, on packagers, and on users. Whether it requires a little patching or a lot of patching doesn't make a difference.
I have yet to see a major new device class, file system class, or other subsystem that didn't require patching. That's a problem with the Linux kernel--it simply lacks the hooks and mechanisms for doing this. And it will only get addressed if the kernel developers start making a commitment to shipping drivers and other modules separately from the main kernel, with their own version numbers and source trees. As long as people can patch easily, they are never going to add the hooks to the kernel that will let new functionality get added without patching.
The problem with Linus getting overloaded is not a problem with SCM, it's a problem with the Linux kernel itself: too many kernel enhancements and bug fixes (apparently) require patches all over the kernel. What we really need is a more flexible way for extensions to hook into the kernel and override existing kernel functionality.
There are lots of ways of providing such hooks. Perhaps the most compatible with the Linux kernel mindset would be something similar to Emacs-hooks: replace most kernel functions with variables holding function pointers to the actual code and provide APIs for manipulating those hooks.
Re:as someone once said...
on
Arguing A.I.
·
· Score: 2
Somewhere, you seem to have lost the thread of the argument. The claim was that supercomputers using existing technology will reach brain-like performance soon. I'm saying: they probably won't.
OTOH, if you want to build a brain-like system out of brain-like stuff, you don't need to wait: just find yourself a partner of the appropriate gender and have a go at it. In about nine months, you'll have the real thing.
ISS is a necessary step to getting to other planets.
Only if you want to send people to other planets. I don't see any reason for doing that any time soon. Automated probes are much more cost effective and will yield enormous amounts of scientific information at much lower cost for many decades to come.
AFAIK, there are no fission or fusion reactors being used in space right now, and no spacecraft carries enough fissionable material for a chain reaction. Putting a reactor into space changes all that and crosses a threshold that we may not want to cross. In fact, I suspect that is the main motivation behind this Bush initiative: once there are reactors in space for civilian and scientific use, it's going to be much harder to object to the use of large quantities of fissionable material in space for military applications.
That's a nice change--I hadn't noticed. It only happened in a recent version, though (1.10, we are now at 1.11). I think that makes CVS even more attractive.
The feature list sounds nice, and using the file system in the way it does is also pretty nice. But I just can't deal with 40kloc of shell script for a version control system. How am I supposed to run that sort of system on a non-UNIX system? What kinds of oddball dependencies is it going to have on the shell, path, and environment?
This seems like it's worse than CVS. Functionally, I'm quite happy with CVS. The main complaint I have about it is that it isn't self-contained but invokes rcs and other shell commands in mysterious ways. "arch" seems to make things worse, not better in that regard. What I would like to see is something mostly like CVS, but something that is implemented as a clean, self-contained library with a single command line executable (with subcommands) and a built-in HTTP-based server. Until that comes along, I think I'll just stick with CVS.
It's good to see that ISS and shuttle funding is going down, although the cuts are not aggressive enough. Those projects provide little scientific benefit and are enormously expensive. They should be wound down quickly.
The missions to the outer planets, I think, are very important and should receive full funding. They may not be very efficient, but travel to the outer planets takes so long and is subject to so many constraints that we really need to get these projects going now. It's a shame that they are being cut.
Nuclear propulsion in space is a hot potato because it's potentially dual-use. If this research is conducted completely openly and in an international framework, then it may be acceptable. Otherwise, it will be perceived as simply a way for the US to militarize space and put nuclear technology into space, and, domestically, it would be little more than a ploy for transferring NASA funds to military research.
The current Dragonball CPU also allows 32bit addressing, it's just that the PalmOS APIs and PalmOS memory management have lots of 16bit limitations. The PalmOS5 announcement mentions nothing about 32bit versions of the API calls, which to me is a major omission. If Palm doesn't manage to make PalmOS5 a full 32bit protected mode platform, they are in serious trouble.
The desktop wars are over and we won't be using a Windows-like desktop in 10-20 years. But we won't be doing all our computing on TiVo's and PDA's either. There is a need for something where people can do knowledge-work: data analysis, writing, programming, etc. The question is: what is that platform going to look like?
Microsoft is trying in their usual, bumbling way. If RedHat wants to be part of the future, they need something a little more visionary than "replace UNIX" and "don't replace the desktop".
But Palm wants their platform to be more than an address book. They want lots of software to be developed for, and ported to, their platform. And for that, it makes sense that their platform has features that allows existing software to be ported to it easily.
Palm needs a window system, a file system, and 32bit addressing not because you need them on a 160x160 device for running an address book, but because having them makes porting other software to the Palm much easier and because it allows PalmOS to be used with bigger screens and in other applications.
Besides, these days, Palm has the worst of both worlds: because memory cards come with DOS file systems, PalmOS is already dealing with file systems, but it still has its oddball database structure in there as well.
The problem is, the general public always gets suckered in by the frills.
No, the problem is that the existing Palm devices are already nice, no-frills devices. But Palm apparently wants to be in something that's higher margin than a sub-$100 consumer device with a simple OS. At least they keep saying so. And if they want to play in that space, they need to provide the features.
As long as you keep the environment simple enough, who really cares about "32-bit addressing" and "memory protection"? A Palm device is intended to be a "temporary holding place for the information contained primarily in your PC"!
Information like images, speech, etc. require a lot of processing power and memory to encode and decode. Furthermore, the existing libraries are written assuming 32bit architectures, and they are a big pain to port to Palm's 16bit architecture (believe me, I have tried).
Young will have to do better than "Linux will replace UNIX" as a long-term strategy for a $300M business. Linux is already replacing UNIX, widely. The question is: what is next?
Linux is a fine desktop replacement, no worse than Windows or MacOSX. If someone wanted to take on those systems, they needed figure out how to bundle Linux with hardware, attract more developers, and market it. But that isn't even the question.
The real question is: after companies like RedHat have extracted much of the value of Linux and other open source software, where are they going to go? What is their vision for the future? "Replacing X with open source software that magically appears" isn't the answer.
In fact, I doubt that in another 10-20 years, we will even have desktops in the traditional sense, and embedded devices will look very different as well. What kind of vision does Young have for that? Not much, it seems.
Legally, if your beta test agreements are worth anything, your customers don't have a leg to stand on. Practically, your testers and customers are right: their feedback and their ideas is what gives your product much of its value in the long run.
In fact, one of the big annoyances with commercial software (this includes Windows, Qt, etc.) is that they are harvesting lots of ideas from their customers and making money from it: customers end up paying for their own work. That's why it is important that people use free software and write bug reports and suggestions for enhancements: that way, the value of using software goes back to the users themselves, rather than to some business.
There are really three groups you need for a great software product:
Programmers, software designers, UI designers, software engineers, etc. Yes, some of them are better than others, but most products are not so complex that you can't find lots of people willing and able to develop them.
Business people and managers with the vision and persistence to create and stick with the product.
Customers that provide lots of feedback and suggestions for enhancements.
Of those, for any good product, the customers are going to do most of the design and drive the evolution of the product.
the real problems is journalists
on
Arguing A.I.
·
· Score: 2
Researchers publish in peer reviewed journals, and you can bet that their peers put a damper on any kind of exaggerated claims.
The people who publish exaggerated claims about AI are journalists eager for a sensational article. Other journalists eager for a story then tell us how we will all get replaced by robots. And then other journalists make a big controversy out of it to publish even more nonsense. And when after just a decade or two AI (or some other overhyped technology) doesn't deliver, journalists write scathing criticisms. To support these claims, journalists scrape together any kind of nut and off-beat comment they can find.
Journalists should stick to reporting science from published, peer-reviewed articles. The real problem is sensationalism and unfounded speculation, and the people responsible for that are journalists. That means you, too, Katz.
as someone once said...
on
Arguing A.I.
·
· Score: 2
"The brain has 100 million neurons, 1 billion of which are in the cerebellum."
Seriously, the units of computation and memory in the brain are likely not individual neurons but synapses, dendritic trees, and even individual channels. That gives you many more orders of magnitude of computational resources for silicon to catch up with. Furthermore, there is no guarantee that Moore's law will continue to hold. In fact, it seems likely that Moore's law will hit the wall just when it comes to trying to get into the realms where biological systems are computing right now.
It's not clear to me what Palm wants to be. The current Palm devices make nifty little address books and calendaring tools with long battery life. But that's a small margin product, which is probably why Handspring is getting out of the business. For corporate applications or multimedia applications, you need fast processors, large screens, and 32bit addressing (so that people don't have to spend an eternity tuning their code).
From the press releases, it's not clear that Palm is addressing any of these issues. OS5 claims no more functionality than you can get on a Sony right now. What about a real window system? What about a real file system? What about a real database? What about 32bit addressing and memory protection? Support for 320x320 screens and some audio and bluetooth APIs isn't going to hack it.
Altogether, OS5 may be more of an incremental improvement over previous versions, offering mostly features that companies like Handera and Sony have already offered on their Palm devices. That may simply not be enough to succeed in the market, given that it's competing against both WinCE and Linux on some nifty hardware.
Well, at least, ARM-based Palms may end up being a nice platform to port Linux to, and it may become available at fire-sale prices if things keep on going this way.
A laptop with 14" LCD, 256M, 20G, DVD, 1GHz, PCMCIA, and USB will cost you around $1000 and it comes with its own built-in UPS. I think, at best, after you have added all the bits and pieces, you'll be able to pull even with the SV24, and it will be a lot more work.
If your goal is merely frame interpolation, I suspect that using a decent MPEG2 encoder and interpolating based on the motion compensation would be good enough.
For other applications (e.g., colorization), you need somewhat better segmentation. Doing this well in the general case is still a research topic; but that's good: you can get lots of research software from around the net that does this sort of thing. Look for keywords like "computer vision", "motion", "segmentation", and "tracking" on Google.
The problem isn't superbowl ads or Britney Spears videos or even your favorite XXX flick. It's that if this information is collected at all, it can be used against you. Insurance agencies could use viewing habits to put you into new risk categories, and overzealous DA's and policemen could use it to pressure you into making a deal or to make you look bad in front of a jury. What's particularly bad about that is that you, as an individual, don't even know what criteria are being used to determine that you are suspicious or a risk.
Why do you want TiVo to succeed in particular? TiVo may have a decent product, and they were first to market. But they weren't the first to have the idea and they don't build anything that half a dozen other companies couldn't build just as well.
I don't see why we need TiVo. If we got standard broadcast and Internet formats for programming information, together with widespread production of these devices by electronics companies, costs would go down and privacy would end up being better.
Well, it's all nice as long as it's all about "girls with bouncy breasts". But what if your government decides to classify you as a terrorist sympathizer because, unlike the rest of the apathetic population, you watched some alternative news coverage? What if your insurance company decides you are too much of a risk because you like to watch car races, watch porn of some kind or other, or watch too many MacDonald's commercials?
Slashdot Mirror
No. It is the nature of big, monolithic programs that don't use modern abstraction facilities to require patches all over. Kernels are no different from word processors in that regard.
That said, your suggestion of using hooks will also require patches 'all over the kernel'.
No. You put in the hooks first, everywhere. Do it from 2.5.10 to 2.5.11. Afterwards, any module can hook just about any function anywhere in the kernel without patches.
I have yet to see a major new device class, file system class, or other subsystem that didn't require patching. That's a problem with the Linux kernel--it simply lacks the hooks and mechanisms for doing this. And it will only get addressed if the kernel developers start making a commitment to shipping drivers and other modules separately from the main kernel, with their own version numbers and source trees. As long as people can patch easily, they are never going to add the hooks to the kernel that will let new functionality get added without patching.
There are lots of ways of providing such hooks. Perhaps the most compatible with the Linux kernel mindset would be something similar to Emacs-hooks: replace most kernel functions with variables holding function pointers to the actual code and provide APIs for manipulating those hooks.
#!/bin/sh
$|=1;
print "foo"; sleep 5; print "bar";
OTOH, if you want to build a brain-like system out of brain-like stuff, you don't need to wait: just find yourself a partner of the appropriate gender and have a go at it. In about nine months, you'll have the real thing.
Only if you want to send people to other planets. I don't see any reason for doing that any time soon. Automated probes are much more cost effective and will yield enormous amounts of scientific information at much lower cost for many decades to come.
AFAIK, there are no fission or fusion reactors being used in space right now, and no spacecraft carries enough fissionable material for a chain reaction. Putting a reactor into space changes all that and crosses a threshold that we may not want to cross. In fact, I suspect that is the main motivation behind this Bush initiative: once there are reactors in space for civilian and scientific use, it's going to be much harder to object to the use of large quantities of fissionable material in space for military applications.
That's a nice change--I hadn't noticed. It only happened in a recent version, though (1.10, we are now at 1.11). I think that makes CVS even more attractive.
This seems like it's worse than CVS. Functionally, I'm quite happy with CVS. The main complaint I have about it is that it isn't self-contained but invokes rcs and other shell commands in mysterious ways. "arch" seems to make things worse, not better in that regard. What I would like to see is something mostly like CVS, but something that is implemented as a clean, self-contained library with a single command line executable (with subcommands) and a built-in HTTP-based server. Until that comes along, I think I'll just stick with CVS.
The missions to the outer planets, I think, are very important and should receive full funding. They may not be very efficient, but travel to the outer planets takes so long and is subject to so many constraints that we really need to get these projects going now. It's a shame that they are being cut.
Nuclear propulsion in space is a hot potato because it's potentially dual-use. If this research is conducted completely openly and in an international framework, then it may be acceptable. Otherwise, it will be perceived as simply a way for the US to militarize space and put nuclear technology into space, and, domestically, it would be little more than a ploy for transferring NASA funds to military research.
The current Dragonball CPU also allows 32bit addressing, it's just that the PalmOS APIs and PalmOS memory management have lots of 16bit limitations. The PalmOS5 announcement mentions nothing about 32bit versions of the API calls, which to me is a major omission. If Palm doesn't manage to make PalmOS5 a full 32bit protected mode platform, they are in serious trouble.
Microsoft is trying in their usual, bumbling way. If RedHat wants to be part of the future, they need something a little more visionary than "replace UNIX" and "don't replace the desktop".
Palm needs a window system, a file system, and 32bit addressing not because you need them on a 160x160 device for running an address book, but because having them makes porting other software to the Palm much easier and because it allows PalmOS to be used with bigger screens and in other applications.
Besides, these days, Palm has the worst of both worlds: because memory cards come with DOS file systems, PalmOS is already dealing with file systems, but it still has its oddball database structure in there as well.
No, the problem is that the existing Palm devices are already nice, no-frills devices. But Palm apparently wants to be in something that's higher margin than a sub-$100 consumer device with a simple OS. At least they keep saying so. And if they want to play in that space, they need to provide the features.
As long as you keep the environment simple enough, who really cares about "32-bit addressing" and "memory protection"? A Palm device is intended to be a "temporary holding place for the information contained primarily in your PC"!
Information like images, speech, etc. require a lot of processing power and memory to encode and decode. Furthermore, the existing libraries are written assuming 32bit architectures, and they are a big pain to port to Palm's 16bit architecture (believe me, I have tried).
Linux is a fine desktop replacement, no worse than Windows or MacOSX. If someone wanted to take on those systems, they needed figure out how to bundle Linux with hardware, attract more developers, and market it. But that isn't even the question.
The real question is: after companies like RedHat have extracted much of the value of Linux and other open source software, where are they going to go? What is their vision for the future? "Replacing X with open source software that magically appears" isn't the answer.
In fact, I doubt that in another 10-20 years, we will even have desktops in the traditional sense, and embedded devices will look very different as well. What kind of vision does Young have for that? Not much, it seems.
In fact, one of the big annoyances with commercial software (this includes Windows, Qt, etc.) is that they are harvesting lots of ideas from their customers and making money from it: customers end up paying for their own work. That's why it is important that people use free software and write bug reports and suggestions for enhancements: that way, the value of using software goes back to the users themselves, rather than to some business.
- Programmers, software designers, UI designers, software engineers, etc. Yes, some of them are better than others, but most products are not so complex that you can't find lots of people willing and able to develop them.
- Business people and managers with the vision and persistence to create and stick with the product.
- Customers that provide lots of feedback and suggestions for enhancements.
Of those, for any good product, the customers are going to do most of the design and drive the evolution of the product.The people who publish exaggerated claims about AI are journalists eager for a sensational article. Other journalists eager for a story then tell us how we will all get replaced by robots. And then other journalists make a big controversy out of it to publish even more nonsense. And when after just a decade or two AI (or some other overhyped technology) doesn't deliver, journalists write scathing criticisms. To support these claims, journalists scrape together any kind of nut and off-beat comment they can find.
Journalists should stick to reporting science from published, peer-reviewed articles. The real problem is sensationalism and unfounded speculation, and the people responsible for that are journalists. That means you, too, Katz.
Seriously, the units of computation and memory in the brain are likely not individual neurons but synapses, dendritic trees, and even individual channels. That gives you many more orders of magnitude of computational resources for silicon to catch up with. Furthermore, there is no guarantee that Moore's law will continue to hold. In fact, it seems likely that Moore's law will hit the wall just when it comes to trying to get into the realms where biological systems are computing right now.
From the press releases, it's not clear that Palm is addressing any of these issues. OS5 claims no more functionality than you can get on a Sony right now. What about a real window system? What about a real file system? What about a real database? What about 32bit addressing and memory protection? Support for 320x320 screens and some audio and bluetooth APIs isn't going to hack it.
Altogether, OS5 may be more of an incremental improvement over previous versions, offering mostly features that companies like Handera and Sony have already offered on their Palm devices. That may simply not be enough to succeed in the market, given that it's competing against both WinCE and Linux on some nifty hardware.
Well, at least, ARM-based Palms may end up being a nice platform to port Linux to, and it may become available at fire-sale prices if things keep on going this way.
A laptop with 14" LCD, 256M, 20G, DVD, 1GHz, PCMCIA, and USB will cost you around $1000 and it comes with its own built-in UPS. I think, at best, after you have added all the bits and pieces, you'll be able to pull even with the SV24, and it will be a lot more work.
For other applications (e.g., colorization), you need somewhat better segmentation. Doing this well in the general case is still a research topic; but that's good: you can get lots of research software from around the net that does this sort of thing. Look for keywords like "computer vision", "motion", "segmentation", and "tracking" on Google.
The problem isn't superbowl ads or Britney Spears videos or even your favorite XXX flick. It's that if this information is collected at all, it can be used against you. Insurance agencies could use viewing habits to put you into new risk categories, and overzealous DA's and policemen could use it to pressure you into making a deal or to make you look bad in front of a jury. What's particularly bad about that is that you, as an individual, don't even know what criteria are being used to determine that you are suspicious or a risk.
I don't see why we need TiVo. If we got standard broadcast and Internet formats for programming information, together with widespread production of these devices by electronics companies, costs would go down and privacy would end up being better.
Well, it's all nice as long as it's all about "girls with bouncy breasts". But what if your government decides to classify you as a terrorist sympathizer because, unlike the rest of the apathetic population, you watched some alternative news coverage? What if your insurance company decides you are too much of a risk because you like to watch car races, watch porn of some kind or other, or watch too many MacDonald's commercials?