For the record, the sig in question is "Ask me for a Gmail invite.".
"Time to update your signature =) We've all gone through the hype Google managed to pump during their IPO phase by releasing an ALPHA version of GMail, and never doing a full release (their PR claim it's beta quality but i beg to differ). Especially now that Yahoo Mail has a search very similar in power to that of GMail"
The Gmail interface still is pretty far ahead of the current Yahoo! one. It remains the best webmail service IMHO.
"I see few reasons to desire a GMail invite."
So... that would be why you didn't ask me for one? I'm a bit lost as to why you felt the need to elaborate further on the matter.
"But my initial point is still that I think Intel will trot out some kind of magical technological revolution that will send AMD scrambling to catch up again. And that we, the consumers, will be the major beneficiaries."
AMD and Intel are going to be trading dominance back and forth forever. They've been doing it forever. There was some floating point coprocessor that AMD made for the (IIRC) 386 that was faster and cheaper than Intel's.
This is just the latest iteration, and Intel happens to have dropped the ball more than usual.
I don't see any reason to make a big show about conserving reaction mass. Reaction mass only matters when you have to bring it with you. With an asteroid, it would be possible to simply scoop bits of it into buckets, and then fling the buckets away as fast as possible given the accelerators we can come up with.
I agree, but I think substantial action will only be taken if a country wealthy enough to do something about it will be hit or affected significantly. If it's going to hit, say, in the south atlantic, the US, the EU, China and Russia (the only ones that could possibly do something about it) will say "Meh, let's just ride this one out.". The resulting tsunami wouldn't do that much damage given a quarter century to prepare, at least not to anyone with money to do something about it.
"I've no experience with 2.6. I moved to FreeBSD-5.3 from 2.4 as my desktop, and so far BSD has been exactly as stable as Linux, which is to say that nothing ever breaks, ever."
I've had significant stability problems with kernel 2.6, even on distros that are supposed to have the resources to do QA on their kernels. Well, not so much stability problems as really annoying bugs that randomly get introduced with each revision.
In enterprise settings, where it's actually possible to track such things, that's easier to believe. All the Windows machines are behind the corporate firewall, while the Linux machines are exposed to the world because Linux server easily outnumber Windows servers, at least for world-visible things.
This has nothing to do with home computers, where incompetent Windows users are pitted against equally incompetent Linux users in competition for the title of "most breached OS".
Given that the market value zombie Windows box is about 5 cents, I think we know who's ahead there.
Not that I'm a big fan of Linux security (My Linux box stays behind my OpenBSD firewall.), but comparing it to Windows is pretty funny.
"Personally, I'd much rather see these things with chemical propulsion until something else non-radioactive (solar, fusion, ?) becomes feasible. There's always a chance that something could fail, and if it burns up in the Earth's atmosphere there could be some nasty fallout."
If the launch vehicle crashed, the worst that could happen would be the release of less uranium than coal power plants already release on a regular basis. Uranium just isn't that radioactive.
The reactor would only be activated when the craft was already in orbit, and a catastrophic crash at that point is very unlikely. To crash, the engines would have to deorbit the craft, which would take a long time. I don't know how long, probably days at least.
You should also be aware that probes to the outer solar system require nuclear power anyway, just not in the form of reactors. They use highly radioactive plutonium to generate heat and electricity. A crash with that would be much more damaging because Plutonium is much more readioactive, and it's already radioactive at launch time. Unused Uranium fuel just isn't that dangerous compared to Plutonium.
"The new IBM is much more focused on giving enterprise customers What They Want. And What They Want is a commodity OS (Linux/Windows) on commodity hardware (x86). They want CHEAP, and IBM will give it to them."
While it's true they want cheap, I don't think many people are married to x86 for things like databases. Those can usually be anything. And when it comes out, the POWER5 will have the lead in database performance by a factor of 3. With numbers like that, nobody cares what the CPU is or what the OS is.
I doubt IBM will require that clients be any platform in particular, but there are advantages to having binary compatability between the servers and the clients. Or perhaps more importantly, binary compatability between giant mainframes and smaller servers.
In any case, the more alternatives there are, the better. And if such a partnership exists, I'm sure they'll eventually be able to convince Apple to sell desktops with ECC memory so they can actually have a proper workstation.
"OpenBSD is definitely on top of the pile, but Linux is quite close. The features are there in Linux, and once they get to where they're doing code readings just to catch bugs, they'll be competitive with OpenBSD on security"
As long as things like new memory leaks continue to turn up in 2.6, I somehow doubt that extensive auditing is going to take place or be of much benefit if it did. The situation isn't going to improve until 2.7 forks, which isn't in the forseeable future.
Also, I honestly don't think the GNU people are willing to do what it takes to match OpenBSD for userspace security. Some of the modifications require cooperations from the kernel and userspace, which means they're not portable. The GNU people take (and deserve) pride in their portability, so it will be a while I think.
And OpenBSD isn't exactly standing still.
"and FreeBSD on speed, features, and stability. Linux is competing well with all three of the old BSDs at once, on their terms."
I do not deny that Linux is a fine OS, but it will be a long time before the stability of 2.6 is comperable to any of the BSDs, and it continues to lack features such as a firewall comparable to PF, something equivilant to the jail facility on {Free,DragonFly}BSD, etc.
"It's hard to kill such a large project... FreeBSD might not die, but there are levels of existance far below where they are now."
My point was just that projects with technical merit have been known to die.
They're trying for a decapitation attack. It's not going to work long term (any more than shutting Napster down did), but I can see how they'd feel they had to do something.
Of course, the problem with doing this is a lot like the problem with antibiotics. If you use them too much, the target adapts.
"there are distributions of Linux right now that rival the BSDs' strong points--except for DragonFlyBSD's."
No... Nobody rivals OpenBSD in terms of security features, and the only one that comes close at the moment is NetBSD. Therefore, there are strong points that Linux does not rival.
"Portage is better than the ports system, and other distributions have binary packages pretty well covered (looking at you, Slackware). At this point, about the only reasons one could claim for choosing FreeBSD over Gentoo are the use of PF, the kernel architecture, or personal preference."
I'm sorry, but that's just wrong.
Portage might be better than the ports tree if someone actually did QA on it. They do not. For example, KDE 3.2 went live with a masked dependency, causing the build to fail. If any of the developers had tried it on a stable system, this problem would have been found and fixed easily. Because no one bothered to try it on a stable system it was broken for a week.
Due to that case and others like it, I have concluded that the Gentoo developers do not do significant QA. That makes it unsuitable for production systems. I for one will not bet my livelihood on someone on the Gentoo forums coming up with a hack to fix some problem before a deadline.
FreeBSD has its problems and it might not survive, but let's not pretend Portage is currently a viable alternative to ports.
The real victory is keeping the variety high enough for websites to be kept from specializing on any given browser. As long as alternatives total more than about 10%, most sites can't afford to require that people use IE.
"Note that NetBSD disables *ALL* services by default, but OpenBSD opens some services by default. Thus, obviously NetBSD is more secure by default."
The only externally accessible service running by default is ssh.
"You can use gcc-ssp (newer version of proplice) on pkgsrc for any daemons you'd like"
OpenBSD automatically uses stack protection for everything, including the kernel. As I said, this has mitigated vulnerabilities in otherwise identical code found on both NetBSD and OpenBSD.
"You are just wrong. NetBSD 2.0 already has this. See below. NetBSD supports PROT_EXEC permission"
I stand corrected on this. I'd checked but obviously missed it.
Slashdot Mirror
For the record, the sig in question is "Ask me for a Gmail invite.".
"Time to update your signature =) We've all gone through the hype Google managed to pump during their IPO phase by releasing an ALPHA version of GMail, and never doing a full release (their PR claim it's beta quality but i beg to differ). Especially now that Yahoo Mail has a search very similar in power to that of GMail"
The Gmail interface still is pretty far ahead of the current Yahoo! one. It remains the best webmail service IMHO.
"I see few reasons to desire a GMail invite."
So... that would be why you didn't ask me for one? I'm a bit lost as to why you felt the need to elaborate further on the matter.
"Wake me when AMD provides complete solutions, chipset, motherboard, with integrated audio and video."
While Intel chipsets have an excellent reputation for stability (one that I can personally support), there are comparable solutions for AMD chips.
"AMD is only a threat to but one small fraction of Intel's business."
Intel's server business is vital to the company.
"But my initial point is still that I think Intel will trot out some kind of magical technological revolution that will send AMD scrambling to catch up again. And that we, the consumers, will be the major beneficiaries."
AMD and Intel are going to be trading dominance back and forth forever. They've been doing it forever. There was some floating point coprocessor that AMD made for the (IIRC) 386 that was faster and cheaper than Intel's.
This is just the latest iteration, and Intel happens to have dropped the ball more than usual.
I don't see any reason to make a big show about conserving reaction mass. Reaction mass only matters when you have to bring it with you. With an asteroid, it would be possible to simply scoop bits of it into buckets, and then fling the buckets away as fast as possible given the accelerators we can come up with.
I agree, but I think substantial action will only be taken if a country wealthy enough to do something about it will be hit or affected significantly. If it's going to hit, say, in the south atlantic, the US, the EU, China and Russia (the only ones that could possibly do something about it) will say "Meh, let's just ride this one out.". The resulting tsunami wouldn't do that much damage given a quarter century to prepare, at least not to anyone with money to do something about it.
"How big is the asteriod?"
I've seen estimates from 300 to 500 meters. It's no planetkiller.
"I've no experience with 2.6. I moved to FreeBSD-5.3 from 2.4 as my desktop, and so far BSD has been exactly as stable as Linux, which is to say that nothing ever breaks, ever."
I've had significant stability problems with kernel 2.6, even on distros that are supposed to have the resources to do QA on their kernels. Well, not so much stability problems as really annoying bugs that randomly get introduced with each revision.
The point of this is to demonstrate that there are small donators as well as big ones.
In enterprise settings, where it's actually possible to track such things, that's easier to believe. All the Windows machines are behind the corporate firewall, while the Linux machines are exposed to the world because Linux server easily outnumber Windows servers, at least for world-visible things.
This has nothing to do with home computers, where incompetent Windows users are pitted against equally incompetent Linux users in competition for the title of "most breached OS".
Given that the market value zombie Windows box is about 5 cents, I think we know who's ahead there.
Not that I'm a big fan of Linux security (My Linux box stays behind my OpenBSD firewall.), but comparing it to Windows is pretty funny.
"Personally, I'd much rather see these things with chemical propulsion until something else non-radioactive (solar, fusion, ?) becomes feasible. There's always a chance that something could fail, and if it burns up in the Earth's atmosphere there could be some nasty fallout."
If the launch vehicle crashed, the worst that could happen would be the release of less uranium than coal power plants already release on a regular basis. Uranium just isn't that radioactive.
The reactor would only be activated when the craft was already in orbit, and a catastrophic crash at that point is very unlikely. To crash, the engines would have to deorbit the craft, which would take a long time. I don't know how long, probably days at least.
You should also be aware that probes to the outer solar system require nuclear power anyway, just not in the form of reactors. They use highly radioactive plutonium to generate heat and electricity. A crash with that would be much more damaging because Plutonium is much more readioactive, and it's already radioactive at launch time. Unused Uranium fuel just isn't that dangerous compared to Plutonium.
"Dont get me wrong, Cassini & Huygens are brilliant, I just wish we had invested more effort into making this sort of mission fundamentally easier."
The Jupiter Icy Moons Oribter will have a nuclear reactor. A proper one, not one of those wimpy radiothermal jobs.
They know the course of the body to some limited precision. The odds represent what portion of possible values would result in a collision with Earth.
"The new IBM is much more focused on giving enterprise customers What They Want. And What They Want is a commodity OS (Linux/Windows) on commodity hardware (x86). They want CHEAP, and IBM will give it to them."
While it's true they want cheap, I don't think many people are married to x86 for things like databases. Those can usually be anything. And when it comes out, the POWER5 will have the lead in database performance by a factor of 3. With numbers like that, nobody cares what the CPU is or what the OS is.
I doubt IBM will require that clients be any platform in particular, but there are advantages to having binary compatability between the servers and the clients. Or perhaps more importantly, binary compatability between giant mainframes and smaller servers.
In any case, the more alternatives there are, the better. And if such a partnership exists, I'm sure they'll eventually be able to convince Apple to sell desktops with ECC memory so they can actually have a proper workstation.
Funny you should mention Hurd...
Hurd and Darwin (the core of MacOS) are both based on Mach.
"OpenBSD is definitely on top of the pile, but Linux is quite close. The features are there in Linux, and once they get to where they're doing code readings just to catch bugs, they'll be competitive with OpenBSD on security"
... FreeBSD might not die, but there are levels of existance far below where they are now."
As long as things like new memory leaks continue to turn up in 2.6, I somehow doubt that extensive auditing is going to take place or be of much benefit if it did. The situation isn't going to improve until 2.7 forks, which isn't in the forseeable future.
Also, I honestly don't think the GNU people are willing to do what it takes to match OpenBSD for userspace security. Some of the modifications require cooperations from the kernel and userspace, which means they're not portable. The GNU people take (and deserve) pride in their portability, so it will be a while I think.
And OpenBSD isn't exactly standing still.
"and FreeBSD on speed, features, and stability. Linux is competing well with all three of the old BSDs at once, on their terms."
I do not deny that Linux is a fine OS, but it will be a long time before the stability of 2.6 is comperable to any of the BSDs, and it continues to lack features such as a firewall comparable to PF, something equivilant to the jail facility on {Free,DragonFly}BSD, etc.
"It's hard to kill such a large project
My point was just that projects with technical merit have been known to die.
meh
If the majority of seasons are in this century, I'm inclined to count it.
"imagine what quality decisions you can make if you simply look for "systems that don't suck."
Well, the decisions may not be of higher quality, but the decision tree is that much more robust."
There's a system that doesn't suck?
News to me.
They were in production in 2001 and later.
There have been exactly 3 good scifi shows this century: Firefly, Farscape, and Stargate SG-1
Can anyone honestly say they liked 5?
They're trying for a decapitation attack. It's not going to work long term (any more than shutting Napster down did), but I can see how they'd feel they had to do something.
Of course, the problem with doing this is a lot like the problem with antibiotics. If you use them too much, the target adapts.
"there are distributions of Linux right now that rival the BSDs' strong points--except for DragonFlyBSD's."
No... Nobody rivals OpenBSD in terms of security features, and the only one that comes close at the moment is NetBSD. Therefore, there are strong points that Linux does not rival.
"Portage is better than the ports system, and other distributions have binary packages pretty well covered (looking at you, Slackware). At this point, about the only reasons one could claim for choosing FreeBSD over Gentoo are the use of PF, the kernel architecture, or personal preference."
I'm sorry, but that's just wrong.
Portage might be better than the ports tree if someone actually did QA on it. They do not. For example, KDE 3.2 went live with a masked dependency, causing the build to fail. If any of the developers had tried it on a stable system, this problem would have been found and fixed easily. Because no one bothered to try it on a stable system it was broken for a week.
Due to that case and others like it, I have concluded that the Gentoo developers do not do significant QA. That makes it unsuitable for production systems. I for one will not bet my livelihood on someone on the Gentoo forums coming up with a hack to fix some problem before a deadline.
FreeBSD has its problems and it might not survive, but let's not pretend Portage is currently a viable alternative to ports.
I would be very nervous about using an unproven server for security reasons.
The real victory is keeping the variety high enough for websites to be kept from specializing on any given browser. As long as alternatives total more than about 10%, most sites can't afford to require that people use IE.
"Note that NetBSD disables *ALL* services by default, but OpenBSD opens some services by default. Thus, obviously NetBSD is more secure by default."
The only externally accessible service running by default is ssh.
"You can use gcc-ssp (newer version of proplice) on pkgsrc for any daemons you'd like"
OpenBSD automatically uses stack protection for everything, including the kernel. As I said, this has mitigated vulnerabilities in otherwise identical code found on both NetBSD and OpenBSD.
"You are just wrong. NetBSD 2.0 already has this. See below.
NetBSD supports PROT_EXEC permission"
I stand corrected on this. I'd checked but obviously missed it.