It is true that microprocessor hardware interfaces have become complicated due to all of the hardware tricks that are done for power management, virtualization, and performance optimizations.
QNX Neutrino, however, seems to run on modern CPUs while remaining a microkernel. I haven't taken a close look at it.
Having filesystem drivers take near-arbitrary strings isn't a bad idea. Making them all URLs is too constraining, though, Because the first element of a URL is a transfer protocol rather than any designation of the data source/sink - that comes later which is sort of backwards. Consider, for example, how the ISO model is layered. They probably are thinking of overlaying other things on the transfer protocol. Not so clean, IMO. A real URL has the implication of running on an IP-like network before the first character. The space of things that filesystems and device drivers can access is much larger.
Well, I read the code. Which isn't a microkernel yet, as far as I can tell. To back that up, note their comment on wishing to move more of the kernel into user space.
I didn't take much time, but the first thing I found in there was a disk driver. A real microkernel would have that in user mode. So, maybe it's an incipient microkernel.
You mean 3c509. I remember a driver developer who would pay you $1 to throw your 3com card away and send him the BNC T-connector. This was in the thin-net days. He contended that this was the only part worth anything.
It's impressive that they are out to make both their own kernel and their own runtime. At first glance it looks like a monolithic kernel, someone should page Andy Tannenbaum to harass them, and if it's really monolithic that takes a point from the "not making other people's mistakes" column.
It takes a lot of computer science smarts to even understand what the mistakes in other operating systems were and how to avoid them. And as others have commented, it's easy to point at other's mistakes when your project is in its infancy, much harder when your project is grown up.
I'll really be impressed if they don't map graphics cards into user space. Nothing's ever stable once you do that. That's the biggest mistake in the whole industry. But I bet they don't take fixing that one on.
As I replied to him, BSD claims support for half a dozen RealTek models, so I can't say he's even accurate about RealTek vs. Intel Ethernet. Now, you at least gave a reason that they might be crap. They're closed. Except that some of them weren't closed enough for a driver to appear on BSD. Did someone get documents, or did they just copy the Linux driver?
If you want uptime, don't have a graphics card in the box. Memory mapping hardware devices into user space is always going to make your box unstable. Both BSD and Linux are really good at staying up without a graphics card in the box.
You haven't given any technical explanation of why RealTek hardware isn't worth the effort, and on this page it shows support in BSD for half a dozen model numbers from RealTek. You must be talking about a different model? There could be many reasons the developers haven't spent time on it other than the quality of the design. For example, a simple lack of people.
You could explain, for example, how the DMA engine in the card is broken or that it drops interrupts (if either of those is the case). Just saying it's crap isn't very credible without technical background.
My impression was that nVidia provided a proprietary driver, thus not long-term supportable once they lose interest.
I have a number of Supermicro boards. Actually their driver and firmware support is not great, and I have complained about their boards to them, only to get the answer that they work correctly on Windows. I buy them because they have remote management, but have given up on their remote management implementations and will go with an external solution next time. My latest Supermicro board is a C7X99-OCE-F and as you can see here, they don't claim BSD support at all. On Linux they have some IOMMU issues at the moment.
It would help me believe this if you could point out "crap" hardware, explain why it's crap, and show that BSD developers decided not to support it because it was crap, with reference to mailing list messages.
Otherwise, it might be better to say "the developers support what they had at a point in time and now only buy what is supported.
Forget robot brains. Give crippled humans remote use of android bodies, and make them telepresence devices.
Regarding taking away jobs, for a long time I've had the feeling that it's not ethical to make a human do a job that a machine can do. Unfortunately, our society is poorly fit to that idea, plutocracy is not a path to a post-scarcity society, but to more plutocracy.
If it's anyone's choice, it comes down to the number of jobs that are created in the region. Use of highways and transit systems is an implication of that.
And it happens that we haven't yet dealt with retarded policies that don't just allow, but depend upon, continuous growth.
If this happened in your body, it would be diagnosed as cancer. Put it in a region, and we kid ourselves that it's progress.
Oh dear! I hope you didn't take that seriously! It was ripped from one of those wacko sites that hate Obama.
Oops, I'd better call off the FBI:-)
Unfortunately, there are a lot of folks who are swayed by wackjob stuff like that, including on Slashdot. Especially since a political party and some news media have been feeding them for more than a decade. Sorry for confusing you with one.
If you want to talk about a real one, it was used on Georgi Markov, and it was detected. I think if there was one that left no trace, it would have already been used on Putin and Trump. 79-year-old men die.
A "cruise missle" is a UAV that is capable of traveling a significant distance at a relatively low speed and carries an explosive payload.
This is obviously not limited to something like the AGM-69.
A small and easily affordable amateur construction is capable of autonomously flying some hundreds of miles and delivering a warhead of several sticks of dynamite to a point with a 30-foot accuracy. One could take this further and equip a Cessna 172 or similar. There are many farm grass strips where such a thing could take off without notice, and it could fly at low altitude. Lots of testing could be carried out with a human pilot in place. And in the end, it could deliver 1000 pounds of explosive.
Look into DIY Drones. Ardupilot is sufficient for a lot of what you are asking for. There's a guy who has successfully flown a model plane across the Atlantic twice. He didn't have constant communications, but it had an HF transmitter and GPS, and it kept hams informed of where it was.
Some of the things you ask for aren't really necessary. Shelf life, for example. Constant communications. Hack proofing. It's really just necessary for a bad guy to put it together and send it toward the nearest city with a few sticks of dynamite. Effective terror weapon. Can distribute poison too. And if one fails, the authorities won't necessarily notice.
Have you tried to buy iodine lately? It's difficult to get because it's somehow used for making methamphetamine. This is annoying, because I would have liked to have some more for my emergency supplies, just for purifying water.
It's pretty easily possible for an amateur to put together their own cruise missle, encrypted communications that admit to no theoretical methods to break them if they're used correctly, spread spectrum radio that you can't tell is there, various sorts of jammers for GPS, phones, etc., various bombs and poisons.
Apple is attempting to be socially responsible. The cell phone is a worse instrument for oppression than Orwell ever imagined. I can make your phone record every moment that you are carrying it. I can compress your voice so well that the existing storage is just fine for that. How long do you think it will be before that's happening for governments, if we embark upon this slope?
The problem is that if you attempt to be socially responsible, the government will do its best to damage your business. Or other companies will. So, corporations have to be cowards to survive.
Ultimately, we can't rely on a corporation for hardware that we can trust. It needs to be independently verifiable. Verifying software is possible. Verifying what is in an IC, less so at present time.
People who don't have noise cancelling headphones and only read the ads for them think that they remove voices. I've yet to meet one that does, having owned three Bose ones over the years, a Phillips one, and one from Sharper Image. They remove the sound of the aircraft engines somewhat, but in general it's easier to hear a voice on a plane with them on than off.
Yes. I have a professional PR person who does my press placements and am married to another, and discuss their work quite a lot. Press releases are meant to incite a news venue to create a story. Corporations also make announcements on their web sites. But everyone should know that these things are in general hyperbolic if they don't just plain lie. So, in general any self-respecting person of normal competence working in a press venue will not copy these things, but will write their own story.
Slashdot has also maintained a greater level of competence than that, other than the not-very-often screw-up. In general those screw ups involve new staff who are completely untrained.
One of the goals was not to do the same old stuff, especially where it was stuff that Tio Paco did 19 years ago that doesn't have any relevancy to today. So, quality content from real people was what interested me. Not really being a news integrator at all. And then I ended up with another, more interesting, project.
Read what you wrote. Slashdot is a news aggregator. News is different from press releases. News is carried by news sites. Press releases are carried by a corporation's own web site and by venues like PR newswire.
Slashdot editors have mostly been smart enough to be able to tell one from the other, up until now. If they aren't able to do that, they aren't going to hold the audience either.
A robot is a machine that can do a human's job. Over time, we cease to think of these jobs as human occupations, and thus we cease to think of the devices as robots. Consider these occupations:
Elevator operator
Washerwoman (it's old enough that I won't say "washerperson")
Slashdot Mirror
It is true that microprocessor hardware interfaces have become complicated due to all of the hardware tricks that are done for power management, virtualization, and performance optimizations.
QNX Neutrino, however, seems to run on modern CPUs while remaining a microkernel. I haven't taken a close look at it.
Having filesystem drivers take near-arbitrary strings isn't a bad idea. Making them all URLs is too constraining, though, Because the first element of a URL is a transfer protocol rather than any designation of the data source/sink - that comes later which is sort of backwards. Consider, for example, how the ISO model is layered. They probably are thinking of overlaying other things on the transfer protocol. Not so clean, IMO. A real URL has the implication of running on an IP-like network before the first character. The space of things that filesystems and device drivers can access is much larger.
I'm not sure we can take Hurd as an indictment of all microkernels. There have been other successful ones.
Well, I read the code. Which isn't a microkernel yet, as far as I can tell. To back that up, note their comment on wishing to move more of the kernel into user space.
Hi David,
I didn't take much time, but the first thing I found in there was a disk driver. A real microkernel would have that in user mode. So, maybe it's an incipient microkernel.
You mean 3c509. I remember a driver developer who would pay you $1 to throw your 3com card away and send him the BNC T-connector. This was in the thin-net days. He contended that this was the only part worth anything.
It's impressive that they are out to make both their own kernel and their own runtime. At first glance it looks like a monolithic kernel, someone should page Andy Tannenbaum to harass them, and if it's really monolithic that takes a point from the "not making other people's mistakes" column.
It takes a lot of computer science smarts to even understand what the mistakes in other operating systems were and how to avoid them. And as others have commented, it's easy to point at other's mistakes when your project is in its infancy, much harder when your project is grown up.
I'll really be impressed if they don't map graphics cards into user space. Nothing's ever stable once you do that. That's the biggest mistake in the whole industry. But I bet they don't take fixing that one on.
As I replied to him, BSD claims support for half a dozen RealTek models, so I can't say he's even accurate about RealTek vs. Intel Ethernet. Now, you at least gave a reason that they might be crap. They're closed. Except that some of them weren't closed enough for a driver to appear on BSD. Did someone get documents, or did they just copy the Linux driver?
If you want uptime, don't have a graphics card in the box. Memory mapping hardware devices into user space is always going to make your box unstable. Both BSD and Linux are really good at staying up without a graphics card in the box.
You haven't given any technical explanation of why RealTek hardware isn't worth the effort, and on this page it shows support in BSD for half a dozen model numbers from RealTek. You must be talking about a different model? There could be many reasons the developers haven't spent time on it other than the quality of the design. For example, a simple lack of people.
You could explain, for example, how the DMA engine in the card is broken or that it drops interrupts (if either of those is the case). Just saying it's crap isn't very credible without technical background.
My impression was that nVidia provided a proprietary driver, thus not long-term supportable once they lose interest.
I have a number of Supermicro boards. Actually their driver and firmware support is not great, and I have complained about their boards to them, only to get the answer that they work correctly on Windows. I buy them because they have remote management, but have given up on their remote management implementations and will go with an external solution next time. My latest Supermicro board is a C7X99-OCE-F and as you can see here, they don't claim BSD support at all. On Linux they have some IOMMU issues at the moment.
:-)
It would help me believe this if you could point out "crap" hardware, explain why it's crap, and show that BSD developers decided not to support it because it was crap, with reference to mailing list messages.
Otherwise, it might be better to say "the developers support what they had at a point in time and now only buy what is supported.
Forget robot brains. Give crippled humans remote use of android bodies, and make them telepresence devices.
Regarding taking away jobs, for a long time I've had the feeling that it's not ethical to make a human do a job that a machine can do. Unfortunately, our society is poorly fit to that idea, plutocracy is not a path to a post-scarcity society, but to more plutocracy.
If it's anyone's choice, it comes down to the number of jobs that are created in the region. Use of highways and transit systems is an implication of that.
And it happens that we haven't yet dealt with retarded policies that don't just allow, but depend upon, continuous growth.
If this happened in your body, it would be diagnosed as cancer. Put it in a region, and we kid ourselves that it's progress.
Oops, I'd better call off the FBI :-)
Unfortunately, there are a lot of folks who are swayed by wackjob stuff like that, including on Slashdot. Especially since a political party and some news media have been feeding them for more than a decade. Sorry for confusing you with one.
If you want to talk about a real one, it was used on Georgi Markov, and it was detected. I think if there was one that left no trace, it would have already been used on Putin and Trump. 79-year-old men die.
A "cruise missle" is a UAV that is capable of traveling a significant distance at a relatively low speed and carries an explosive payload.
This is obviously not limited to something like the AGM-69.
A small and easily affordable amateur construction is capable of autonomously flying some hundreds of miles and delivering a warhead of several sticks of dynamite to a point with a 30-foot accuracy. One could take this further and equip a Cessna 172 or similar. There are many farm grass strips where such a thing could take off without notice, and it could fly at low altitude. Lots of testing could be carried out with a human pilot in place. And in the end, it could deliver 1000 pounds of explosive.
Look into DIY Drones. Ardupilot is sufficient for a lot of what you are asking for. There's a guy who has successfully flown a model plane across the Atlantic twice. He didn't have constant communications, but it had an HF transmitter and GPS, and it kept hams informed of where it was.
Some of the things you ask for aren't really necessary. Shelf life, for example. Constant communications. Hack proofing. It's really just necessary for a bad guy to put it together and send it toward the nearest city with a few sticks of dynamite. Effective terror weapon. Can distribute poison too. And if one fails, the authorities won't necessarily notice.
Have you tried to buy iodine lately? It's difficult to get because it's somehow used for making methamphetamine. This is annoying, because I would have liked to have some more for my emergency supplies, just for purifying water.
Yes, ways of taking the encapsulation off are known, and the chip can sometimes be analyzed once you do that. But that only analyzes a sample,
and it does it destructively. To verify a chip, you need to be able to verify the working one in your own device.
It's pretty easily possible for an amateur to put together their own cruise missle, encrypted communications that admit to no theoretical methods to break them if they're used correctly, spread spectrum radio that you can't tell is there, various sorts of jammers for GPS, phones, etc., various bombs and poisons.
Not that I really want to tell this to Congress.
Apple is attempting to be socially responsible. The cell phone is a worse instrument for oppression than Orwell ever imagined. I can make your phone record every moment that you are carrying it. I can compress your voice so well that the existing storage is just fine for that. How long do you think it will be before that's happening for governments, if we embark upon this slope?
The problem is that if you attempt to be socially responsible, the government will do its best to damage your business. Or other companies will. So, corporations have to be cowards to survive.
Ultimately, we can't rely on a corporation for hardware that we can trust. It needs to be independently verifiable. Verifying software is possible. Verifying what is in an IC, less so at present time.
People who don't have noise cancelling headphones and only read the ads for them think that they remove voices. I've yet to meet one that does, having owned three Bose ones over the years, a Phillips one, and one from Sharper Image. They remove the sound of the aircraft engines somewhat, but in general it's easier to hear a voice on a plane with them on than off.
Yes. I have a professional PR person who does my press placements and am married to another, and discuss their work quite a lot. Press releases are meant to incite a news venue to create a story. Corporations also make announcements on their web sites. But everyone should know that these things are in general hyperbolic if they don't just plain lie. So, in general any self-respecting person of normal competence working in a press venue will not copy these things, but will write their own story.
Slashdot has also maintained a greater level of competence than that, other than the not-very-often screw-up. In general those screw ups involve new staff who are completely untrained.
One of the goals was not to do the same old stuff, especially where it was stuff that Tio Paco did 19 years ago that doesn't have any relevancy to today. So, quality content from real people was what interested me. Not really being a news integrator at all. And then I ended up with another, more interesting, project.
Read what you wrote. Slashdot is a news aggregator. News is different from press releases. News is carried by news sites. Press releases are carried by a corporation's own web site and by venues like PR newswire.
Slashdot editors have mostly been smart enough to be able to tell one from the other, up until now. If they aren't able to do that, they aren't going to hold the audience either.
A robot is a machine that can do a human's job. Over time, we cease to think of these jobs as human occupations, and thus we cease to think of the devices as robots. Consider these occupations: