I really like using PF on OpenBSD, and would love to see it on Linux. I think it is much more powerful and easier to configure than any other filtering system for a F/OSS operating system.
Meanwhile, continue to try to commoditize hardware by pushing Java, even though you are a hardware company and that's the last thing in the world you should want.
Yeah, that's something I never understood with Sun's strategy.
On the one hand, they needed/wanted a viable server platform that ran on Solaris. But they now need a viable server platform that runs only on Solaris.
Sun makes spiffy (if expensive) hardware. But as people improve clustering solutions on cheap hardware, Sun will feel greater and greater pain.
Well, I believe the guy is a corprate shill, but that's just my opinion.
He does make one valid point, though. Microsoft is not evil. Any more than an animal out in the forest is evil.
Good and evil imply moral choice. Companies like M$ don't make moral choices, they make business ones. (It could well be argued that since corporations are run by people, that they should make moral choices, but I'll leave that discussion to another day.)
Microsoft is a implacable and ruthless competitor. I don't like their products, and I don't like using them. I don't like being forced into their way of doing things, and their way of thinking.
But that makes them something to be avoided. It doesn't make them evil.
I was talking about multiple ducts because I think cross-linking is a bad idea. I think that the cost in terms of weight and complexity outweights the redundancy advantages. So yes, the idea was the system would have enough excess lift capacity to handle the loss of a single duct.
A military version of the skycar wouldn't have some of the same constraints as the civilian one. Moller is worried about noise and ground damage during take-off. Marines don't care about that. You could increase the size of the ducts, and reduce disc loading a bit, if you think that is a good idea.
And if you have enough power to hover, that means you have plenty of power to go fast in horizontal flight. Which is what I thought was the point. If you want a helicopter, by all means, get a helicopter.
As for fuel efficiency, sure, the skycar has terrible efficiency during hover. So what? Civilian or military, you don't want to spend much time in hover anyway. The point is to go fast from point A to point B, and do VTOL from unprepared surfaces. Hovering just makes you an easier target.
I think the V-22 has had problems just because it is too complex.
You've got two jet-turbines, which can each power both rotors. So you've got a very complex power distribution system. Lots of stuff in those pods which rotate, so lots of flexible connections which can break.
I would have preferred to see a design with six or so smaller ducted fans. So even if you lose one on each side (due to small arms fire, for example) you still have enough power to maneuver and land safely. Two or three lost on one side would need ballistically deployed parachutes to land.
Hmph. I've just described a Moller skycar. The production version hasn't flown yet. But with relatively modest funding, I bet it could. Still got a complex computer control system, so who knows what bugs might lurk there.
We're headed towards the "Paperless Office". The road is longer and bumpier than was first imagined, but we're getting there.
The only times I print out stuff is when it needs to be portable (like printing driving directions) and I don't want to putz with putting it on a PDA.
Or sometimes, flipping through a document is easier than viewing it on the screen. I wish I had a PDF viewer which was really, really fast. Maybe something that could pre-render pages without gobbling massive amounts of memory...
Stuff like printing out code is almost useless. How can I tell if I'm looking at the latest version?
A lot of the notes and stuff I write these days goes into documentation, or the coporate wiki. Writing something down on paper only benefits me. Putting it on the wiki can potentially benefit everyone.
Except, of course, that iD software is one of the few companies that doesn't like to do that.
They tend not to release software until 'it is ready'. That's because they have enough money (and they control their own company) that they set the release schedule.
Private industry can not take up the flag for space exploration until a cheaper way is found to get there. In order for that to happen there must be central agency that focuses on this, and the agency must get government funding.
Bullshit.
Rocket science isn't "Rocket Science" anymore. That was 40 years ago. We know how to do cheap launches. We could put payloads in orbit for 1/100th the cost that NASA does with the shuttle.
The problem is that we've created this bureaucratic mess called NASA. Between them, the major aerospace contractors (who don't want their meal tickets cancelled) and the FAA, they're preventing anyone from developing cheaper and more reliable launch platforms. Read about the trials and tribulations of the X-Prize conteners for example.
What is sorely needed is a public champion to advocate for increased NASA funding, as part of a commitment to keep America at the forefront of technological leadership worldwide.
Well, to be very blunt: Fuck that.
I will say to anyone and everyone that NASA gets far, far too much money, for few tangible results.
Just take a look at what the X-Prize contenders are doing with thousands (not millions and certainly not billions) of dollars. It costs hundreds of millions per shuttle launch (probably more after the accident). Can you imagine what we could accomplish if we gave that money as a prize to organizations demonstrating cheap and safe lauch vehicles?
Through a variety of factors, NASA and their buddies the major aerospace contractors have strangled real research and innovation in space technology. We could have had tourist rides to orbit and space hotels already, for the money that was spent on the shuttle program. And more digital radio and TV, and more location services, and cheap satellite mapping, and so on.
Burning more money on a 30 year old launch vehicle isn't keeping us at the forefront of technology. It is holding us back. NASA these days is a jobs program for the aerospace industry. They do things to keep people busy, rather than helping humanity go to the planets and stars.
The fact is, SCO is still distributing Linux code they say violates their IP. Sure, they stopped selling Linux after they filed the lawsuit (not the day of, like they should have...), but they've still got it on their FTP server.
Even if everything else they say is true about their contracts w/ IBM, the fact is they've done nothing to mitigate damages. Nothing, in all this time.
So if, through some travesty of justice, they win their case against IBM, they have no way to go after anyone else using Linux. The disputed code can stay in Linux because they've allowed it themselves.
Yeah, but if they remove all support for OpenServer and Unixware, then they will push the cost of support for those platforms onto SCO, where it belongs.
It is one thing to allow them to continue using Samba. It is another thing to let them have a free ride.
Make them pay!
Re:dan bernstein's position on this
on
DNSSEC: Good Enough?
·
· Score: 2, Interesting
I've never tried his DNS implementation, but I've heard it works nicely.
It does work nicely. Been using it for about 3 years now. No problems at all. After you get used to the DJB way of doing things, it is very simple to configure. The main data file makes more sense to me than BIND's stuff ever did.
But DJB is out there. One of these days, in my copious free time, I'll have to re-implement some of his better ideas, so that they can be released under a normal F/OSS license.
But I'm not using Qmail any more. Hasn't been updated in years, and to get needed features, it is patch hell. Switched to Courier MTA because I needed mail filtering, webmail and IMAP. I still like Maildirs though. Never had a problem with mailbox corruption or lost messages since we switched to that.
Basically because they messed up, and got too greedy.
They aren't willing to give up their exsiting support contracts for Linux, so they kept distributing the Linux kernel even after they claimed it violated their own precious IP.
So now they're stuck with a situation where the obvious thing to do to mitigate damages is to completely stop distribution. And so they stopped selling it. But they've still got source packages on their FTP site.
Since they still want support contract money (well, considering how much they're spending on lawyers, I'd want it too I guess) the only other way out of this dead-end is to delcare the GPL invalid. That way, they can continue to distribute Linux, while charging fees for its use.
That's one of the reasons why I'm hopeful that things will work out well in the end. They are so greedy they're making stupid moves.
I'm sure things have improved tremendously, but in the previous decade, Apple computers were a severe pain in the posterior to support in a large enviroment.
There's a lot of things about Appletalk that didn't scale well at all. I wasn't a member of the Mac support team, but oh, the stories I could tell... Oh, the hacks that were needed to get them onto the regular TCP/IP network...
If sysadmins aren't installing Macs now, maybe that's why. Maybe they are just afraid.
So how easy are they to integrate into a large network these days?
You know, what's really going to lose the case for SCO is that they just got too greedy.
If they had stopped distributing in any fashion Linux on the day they announced the lawsuit, they might have still had a chance.
But since they are continuing to distribute Linux through their website, they have done nothing to curb infringement of their "valuable IP".
I guess they couldn't live without those support contracts for Linux they had already signed. There's smart greedy, and dumb greedy. And so a few hundred thousand dollars will now cost them 3 billion (potentially). Dumb, dumb, Dumb.
If SBC was so concerned about privacy, they wouldn't have you install so much adware and spyware when you order DSL service from them.
Feh, you wouldn't believe all the crap that gets installed, and with no option for not installing it.
And on top of that, they implicitly encourage people to directly connect their PCs to the internet, without any firewall or firewall software. So I guess they're not concerned when some skript kiddie installs a keylogger on your PC, and transfers money out of your bank account.
Out by an order of magnitude, and not getting closer any time soon. Last I checked, it's out by a factor of 100 or so for small operations (loops, etc)
Well, regular Python code can see some dramatic speedups, with just a little bit of extra work. Check out
Psyco if you firmly believe that interpreted languages will never ever be as fast as C.
With sufficient cleverness, it may be possible to boost Python beyond the speed of the most highly optimized pre-compiled program.
Psyco and approaches like it do have drawbacks, but there is some very interesting work going on now with high-level languages.
Well, not sure what everyone's doing in this space. But its not what Sun is doing.
I looked at Sun's whitepaper: The Java HotSpot Virtual Machine, v1.4.1. When they're talking about hotspots, they just mean running the JIT on the most-used sections of code.
They're trying to get Java bytecode to match the speed of optimized C code.
The Psyco stuff I mentioned is talking about dynamic optimimization of functions you've already selected. They are trying to surpass the speed of optimized C code.
Even if you're not a fan of Python, I'd recommend reading the stuff on the Psyco site, just to get some interesting ideas.
What's interesting about Psyco and similar efforts is that there exists a possibility of further optimzations than in regular compiled languages.
Suppose you have a function called do_it(), which takes an integer argument. Its got a loop in there that does some calculation. If you wrote it in C, you'd code up your for loop, and do it's thing.
However, a really smart optimization system would not only look at the code as you've written it, but also look how it is actually used in runtime.
Supposed that most of the time you're calling do_it() with an argument of '3'. It may be possible to really optimize that code path, and use either the regular version, or the '3' version based on the function's argument.
This is much better than trying to do it by hand. Becuase your next program may mostly call do_it() with '5' instead of '3'.
This kind of dynamic optimization is only possible when you're using a high level language, which embodies the general algorithm. It is nice when you can leave grungy details like optimization up to the system instead, who can (potentially) do a much better job at it than you can. Read the papers on the site, in case I'm not explaining things well.
Slashdot Mirror
I really like using PF on OpenBSD, and would love to see it on Linux. I think it is much more powerful and easier to configure than any other filtering system for a F/OSS operating system.
Please, please, pretty please!
Meanwhile, continue to try to commoditize hardware by pushing Java, even though you are a hardware company and that's the last thing in the world you should want.
Yeah, that's something I never understood with Sun's strategy.
On the one hand, they needed/wanted a viable server platform that ran on Solaris. But they now need a viable server platform that runs only on Solaris.
Sun makes spiffy (if expensive) hardware. But as people improve clustering solutions on cheap hardware, Sun will feel greater and greater pain.
The 2nd version of the patch for DJBDNS, which has instructions inside is at:
http://tinydns.org/djbdns-1.05-ignoreip2.patch
I thought before that Verisign was a bunch of fuckers.
And now they've completely convinced me of that.
Shit. What a bunch of fuckers.
Well, I believe the guy is a corprate shill, but that's just my opinion.
He does make one valid point, though. Microsoft is not evil. Any more than an animal out in the forest is evil.
Good and evil imply moral choice. Companies like M$ don't make moral choices, they make business ones. (It could well be argued that since corporations are run by people, that they should make moral choices, but I'll leave that discussion to another day.)
Microsoft is a implacable and ruthless competitor. I don't like their products, and I don't like using them. I don't like being forced into their way of doing things, and their way of thinking.
But that makes them something to be avoided. It doesn't make them evil.
I was talking about multiple ducts because I think cross-linking is a bad idea. I think that the cost in terms of weight and complexity outweights the redundancy advantages. So yes, the idea was the system would have enough excess lift capacity to handle the loss of a single duct.
A military version of the skycar wouldn't have some of the same constraints as the civilian one. Moller is worried about noise and ground damage during take-off. Marines don't care about that. You could increase the size of the ducts, and reduce disc loading a bit, if you think that is a good idea.
And if you have enough power to hover, that means you have plenty of power to go fast in horizontal flight. Which is what I thought was the point. If you want a helicopter, by all means, get a helicopter.
As for fuel efficiency, sure, the skycar has terrible efficiency during hover. So what? Civilian or military, you don't want to spend much time in hover anyway. The point is to go fast from point A to point B, and do VTOL from unprepared surfaces. Hovering just makes you an easier target.
I think the V-22 has had problems just because it is too complex.
You've got two jet-turbines, which can each power both rotors. So you've got a very complex power distribution system. Lots of stuff in those pods which rotate, so lots of flexible connections which can break.
I would have preferred to see a design with six or so smaller ducted fans. So even if you lose one on each side (due to small arms fire, for example) you still have enough power to maneuver and land safely. Two or three lost on one side would need ballistically deployed parachutes to land.
Hmph. I've just described a Moller skycar. The production version hasn't flown yet. But with relatively modest funding, I bet it could. Still got a complex computer control system, so who knows what bugs might lurk there.
We're headed towards the "Paperless Office". The road is longer and bumpier than was first imagined, but we're getting there.
The only times I print out stuff is when it needs to be portable (like printing driving directions) and I don't want to putz with putting it on a PDA.
Or sometimes, flipping through a document is easier than viewing it on the screen. I wish I had a PDF viewer which was really, really fast. Maybe something that could pre-render pages without gobbling massive amounts of memory...
Stuff like printing out code is almost useless. How can I tell if I'm looking at the latest version?
A lot of the notes and stuff I write these days goes into documentation, or the coporate wiki. Writing something down on paper only benefits me. Putting it on the wiki can potentially benefit everyone.
Except, of course, that iD software is one of the few companies that doesn't like to do that.
They tend not to release software until 'it is ready'. That's because they have enough money (and they control their own company) that they set the release schedule.
Well, for ternary computing, you could do -3.3 V, 0V and 3.3V. So you don't actually need higher absolute voltage levels.
Indeed. The Saturn V was used for moon missions as well as launching a space station in once piece.
It's not the ideal platform for fixing satellites, but if lauch costs were lower, we'd just send up new ones, rather than trying to repair them.
Private industry can not take up the flag for space exploration until a cheaper way is found to get there. In order for that to happen there must be central agency that focuses on this, and the agency must get government funding.
Bullshit.
Rocket science isn't "Rocket Science" anymore. That was 40 years ago. We know how to do cheap launches. We could put payloads in orbit for 1/100th the cost that NASA does with the shuttle.
The problem is that we've created this bureaucratic mess called NASA. Between them, the major aerospace contractors (who don't want their meal tickets cancelled) and the FAA, they're preventing anyone from developing cheaper and more reliable launch platforms. Read about the trials and tribulations of the X-Prize conteners for example.
What is sorely needed is a public champion to advocate for increased NASA funding, as part of a commitment to keep America at the forefront of technological leadership worldwide.
Well, to be very blunt: Fuck that.
I will say to anyone and everyone that NASA gets far, far too much money, for few tangible results.
Just take a look at what the X-Prize contenders are doing with thousands (not millions and certainly not billions) of dollars. It costs hundreds of millions per shuttle launch (probably more after the accident). Can you imagine what we could accomplish if we gave that money as a prize to organizations demonstrating cheap and safe lauch vehicles?
Through a variety of factors, NASA and their buddies the major aerospace contractors have strangled real research and innovation in space technology. We could have had tourist rides to orbit and space hotels already, for the money that was spent on the shuttle program. And more digital radio and TV, and more location services, and cheap satellite mapping, and so on.
Burning more money on a 30 year old launch vehicle isn't keeping us at the forefront of technology. It is holding us back. NASA these days is a jobs program for the aerospace industry. They do things to keep people busy, rather than helping humanity go to the planets and stars.
But all that isn't even the worst bit.
The fact is, SCO is still distributing Linux code they say violates their IP. Sure, they stopped selling Linux after they filed the lawsuit (not the day of, like they should have...), but they've still got it on their FTP server.
Even if everything else they say is true about their contracts w/ IBM, the fact is they've done nothing to mitigate damages. Nothing, in all this time.
So if, through some travesty of justice, they win their case against IBM, they have no way to go after anyone else using Linux. The disputed code can stay in Linux because they've allowed it themselves.
Yeah, but if they remove all support for OpenServer and Unixware, then they will push the cost of support for those platforms onto SCO, where it belongs.
It is one thing to allow them to continue using Samba. It is another thing to let them have a free ride.
Make them pay!
I've never tried his DNS implementation, but I've heard it works nicely.
It does work nicely. Been using it for about 3 years now. No problems at all. After you get used to the DJB way of doing things, it is very simple to configure. The main data file makes more sense to me than BIND's stuff ever did.
But DJB is out there. One of these days, in my copious free time, I'll have to re-implement some of his better ideas, so that they can be released under a normal F/OSS license.
But I'm not using Qmail any more. Hasn't been updated in years, and to get needed features, it is patch hell. Switched to Courier MTA because I needed mail filtering, webmail and IMAP. I still like Maildirs though. Never had a problem with mailbox corruption or lost messages since we switched to that.
Basically because they messed up, and got too greedy.
They aren't willing to give up their exsiting support contracts for Linux, so they kept distributing the Linux kernel even after they claimed it violated their own precious IP.
So now they're stuck with a situation where the obvious thing to do to mitigate damages is to completely stop distribution. And so they stopped selling it. But they've still got source packages on their FTP site.
Since they still want support contract money (well, considering how much they're spending on lawyers, I'd want it too I guess) the only other way out of this dead-end is to delcare the GPL invalid. That way, they can continue to distribute Linux, while charging fees for its use.
That's one of the reasons why I'm hopeful that things will work out well in the end. They are so greedy they're making stupid moves.
I'm sure things have improved tremendously, but in the previous decade, Apple computers were a severe pain in the posterior to support in a large enviroment.
There's a lot of things about Appletalk that didn't scale well at all. I wasn't a member of the Mac support team, but oh, the stories I could tell... Oh, the hacks that were needed to get them onto the regular TCP/IP network...
If sysadmins aren't installing Macs now, maybe that's why. Maybe they are just afraid.
So how easy are they to integrate into a large network these days?
You know, what's really going to lose the case for SCO is that they just got too greedy.
If they had stopped distributing in any fashion Linux on the day they announced the lawsuit, they might have still had a chance.
But since they are continuing to distribute Linux through their website, they have done nothing to curb infringement of their "valuable IP".
I guess they couldn't live without those support contracts for Linux they had already signed. There's smart greedy, and dumb greedy. And so a few hundred thousand dollars will now cost them 3 billion (potentially). Dumb, dumb, Dumb.
I'm going to send them $731 dollars for my TiVo and Linux system.
I'm paying them with Monopoly money though, as that's about what their IP is worth.
If SBC was so concerned about privacy, they wouldn't have you install so much adware and spyware when you order DSL service from them.
Feh, you wouldn't believe all the crap that gets installed, and with no option for not installing it.
And on top of that, they implicitly encourage people to directly connect their PCs to the internet, without any firewall or firewall software. So I guess they're not concerned when some skript kiddie installs a keylogger on your PC, and transfers money out of your bank account.
Out by an order of magnitude, and not getting closer any time soon. Last I checked, it's out by a factor of 100 or so for small operations (loops, etc)
Well, regular Python code can see some dramatic speedups, with just a little bit of extra work. Check out Psyco if you firmly believe that interpreted languages will never ever be as fast as C.
With sufficient cleverness, it may be possible to boost Python beyond the speed of the most highly optimized pre-compiled program.
Psyco and approaches like it do have drawbacks, but there is some very interesting work going on now with high-level languages.
Well, the GPL is intended to empower software users. It is not as empowering to software developers.
Well, not sure what everyone's doing in this space. But its not what Sun is doing.
I looked at Sun's whitepaper: The Java HotSpot Virtual Machine, v1.4.1. When they're talking about hotspots, they just mean running the JIT on the most-used sections of code. They're trying to get Java bytecode to match the speed of optimized C code.
The Psyco stuff I mentioned is talking about dynamic optimimization of functions you've already selected. They are trying to surpass the speed of optimized C code.
Even if you're not a fan of Python, I'd recommend reading the stuff on the Psyco site, just to get some interesting ideas.
What's interesting about Psyco and similar efforts is that there exists a possibility of further optimzations than in regular compiled languages.
Suppose you have a function called do_it(), which takes an integer argument. Its got a loop in there that does some calculation. If you wrote it in C, you'd code up your for loop, and do it's thing.
However, a really smart optimization system would not only look at the code as you've written it, but also look how it is actually used in runtime.
Supposed that most of the time you're calling do_it() with an argument of '3'. It may be possible to really optimize that code path, and use either the regular version, or the '3' version based on the function's argument.
This is much better than trying to do it by hand. Becuase your next program may mostly call do_it() with '5' instead of '3'.
This kind of dynamic optimization is only possible when you're using a high level language, which embodies the general algorithm. It is nice when you can leave grungy details like optimization up to the system instead, who can (potentially) do a much better job at it than you can. Read the papers on the site, in case I'm not explaining things well.