And not to go too far offtopic, but this is the reason why any human designed economy like socialism (currently called liberalism), is doomed to fail.
Capitalism has failings, but in general things work themselves out in the end. Monopolies collapse under their own weight. Abusive employers lose good employees which hurts them in the long run. Things like that.
I thought the whole premise of the hate crime laws was that the offender had to be doing something else illegal that was motivated by hate, and that was the only way it skirted the 1st amendment.
I guess all the people that say "slippery slope" is an invalid argument don't know much about governments or laws.
That was his point. The segregation isn't imposed by the majority, the segregation is imposed and enforced by the minorities themselves, because they don't really want to end racism and segregation either.
Because not everyone is getting software from the Internet. When the GPL was written, the Internet as we know it today didn't even exist.
The clause still makes sense in the Internet world though. Suppose I include a link to a third party site to get GPL code I distribute. They go out of business or change their domain name, or even take down the version I used because it became obselete. I'm now in violation of the GPL.
So I have a good uptime, until I try to upgrade a kernel, which requires me to dick with a lilo.conf and rebuild an initrc by hand. That's still downtime.
I don't have a problem with doing that, it's just unexpected, since Red Hat has had kernel upgrades be fully automatic for almost 5 years now.
There's a tradeoff between time spent fixing stability problems, and time spent doing mundane tasks that should be automatic. The mundane tasks are much more common.
Because there's a lot of untapped value there. When Debian is still using ghostscript 6, and thousands of other obselete packages, you have to sacrifice all the advances made in the last several years if you want to run Debian.
Testing/Sid isn't an option for production, since the Debian people won't commit to providing security fixes in any timely manner for those versions.
I think Red Hat used to have it right. A distro should be updated in a major way about once every year-18 months. Ideally you'd want to support the current version, and the immediate previous major version.
It worked great, I don't know why RH had to fuck it up.
Books and learning with people that already know are sufficient. The lecture model is obselete.
Re:You talk to the developer
on
Debugging Configure
·
· Score: 2, Interesting
Well the danger with that, is the authors being deluged by people who just don't have the right dependancies installed.
Of course, it could be considered a bug if the configure script fails silently or ambiguously just because a dependancy was missing, but I see it a lot.
It'd really help if the authors would put very verbose missing dependancy messages, even as far as including a URL for the dependancies, if they aren't common.
One build that comes to mind that was total hell was Flightgear. That thing has dependancies all over the place, and not common things either.
My son fell down at an after school program, and broke his finger. We took him to the local walk-in clinic, and the doctor there took x-rays. She said she thought it should be referred to a specialist because the fracture was close to the joint.
She put a splint on it.
Took him to the specialist, they took x-rays, said it would be fine, and put a splint on it. Took maybe 5 minutes of the doctor's time, 5 minutes of the x-ray tech's time.
Get the bill, and the specialist charged my insurance company for $450 for "surgical services". That's not including the charge for the x-ray.
Called my insurance company (since it says right on the notice, to report suspected fraud). Insurance company says I need to call the specialist. Called the specialist and they said that's the normal way they bill "fracture care services".
So I guess there's nothing I can do. I had to pay 10% or something as a copayment, so it was like $40 for me, in addition to two other $20 copayments for the specialists x-ray and the walk-in clinic doctor.
Obviously I don't care about such small amounts of money, but frankly I'm stunned the insurance company would pay $450 for a fucking 5 minute glance at an xray, something the GP already did.
There's obviously corruption, bribes, kickbacks, all over the place in the medical insurance industry. If medical insurance is in "crisis", then it's because of this.
Just like the "energy crisis" with Enron, the "medical crisis" is a pile of shit. It's a coverup for the lies and corruption throughout the medical industry. In 5 or 10 years maybe we'll see some major scandal about medical billing. Until then the criminals behind this will continue to fleece everyone.
If the people who were intelligent enough to not need university were not discriminated against, then a lot of people wouldn't waste time and money on university, developing unneeded debts.
The false idea that university education is somehow necessary for so many jobs is costing our society more than we realize.
rdiff-backup and rsync-incremental backup shell scripts do something very similar, though it's only a smear-shot if the filesystem is changing.
We're using it to successfully do daily backups for the last 30 days for several TBs with very little overhead. No need for periodic full backups that would take days, since the diffs go back in time instead of forward.
XFS and ext3 I've both taken up to 50,000+ files in the same directory. There's no noticable performance lag on any of them. XFS does seem marginally better, but ext3 is about on-par. Once XFS is in the stock kernel for distros, I'll probably use it a lot more, it does seem to handle very large filesystems a little better.
I've always had trouble setting RAID stride for ext3 to actually see any performance increase, and the ext3 userland utils suck ass, with confusing documentation, and really bad defaults for large file systems. (The defaults don't scale)
I think Reiser's claims are overrated. It's nice to have a competitor, and I'm not anti-Reiser or anything, I just haven't seen any benefit, and I've run into serious bugs in Reiser in the past, even after all the Reiser fans were proclaiming stability.
And finally, the truth is, most people I know don't even delete files
I'm not discussing a file system for those people, I have storage and network file servers in mind.
Well, I hate defending MS, but it does sorta make sense to swap out stuff that hasn't been used in a while, even if there's lots of free RAM. Then if you start something that really needs tons of RAM, you won't have to wait for swap then, of stuff you weren't going to use anyway.
It's best if the algorithm has some idea what "big" is, but that's the difficulty in designing a VM system that works well with 128megs of ram or less and also 4+GB.
Gun ownership is a responsibility, not just a "basic human right."
Rights always require responsibility. Liberty requres plenty of responsibility, and yet you consider it a basic human right.
Our Constitution gives us this right
The Consistiution doesn't give rights. It merely enumerates what the founders thought were the most important inherent rights to form a long lasting stable state that offers freedom.
Note that later in the bill of rights they mention "rights retained by the people". This should be clear enough evidence that the bill of rights should not be construed to be granting rights, only to enumerate rights that already exist.
for one explicit purpose -- to maintain a militia
This is a dead horse debate over wording, but the way the 2nd is written, it's clearly not meant to be the sole purpose. If I say "Money is good, I should go get a job", that doesn't mean I want to get a job solely because "money is good", it's simply one reason that I found compelling.
In other words, owning a firearm is part of every man's civic duty to protect his country and resist oppresion and is not an inherent right to go waving about like you automatically deserve to handle a dangerous weapon simply for breathing.
I think you misunderstand the term "right". I think you've internalized the liberal definition of "human rights", which is definitely streched from what the founders intended. They made it clear that rights require a certain level of responsibility in exercising of the rights. They didn't have a problem with punishing people who abused thier rights.
The supreme court has been generally been very careful to avoid "prior restraint" in 1st amendment issues. The idea is that everyone should have the ability to exercise their rights, and they shouldn't be punished until and unless they abuse their rights.
As far as whether the line is arbitrary or not, you mention cost/benefit ratios. That's not how constitutional law has ever been interpreted. There's obviously a high cost in social stability in allowing the KKK to march, and very little social benefit. Cost/benefit has never been a factor when it comes to fundamental rights. If it ever becomes a major factor, we are in serious trouble.
In essence, the 2nd Amendment is unfortunately obsolete.
Well, the founders couldn't have possibly forseen the Internet, so the 1st amendment must be obselete too. Hell, why don't we repeal all 10 amendments in the bill of rights, they are pretty much all "obselete" under your argument.
The 2nd amendment doesn't say "as long as it's easy", or "until better weapons are invented". Your argument flies in the face of the Bill of Rights as a whole.
Until someone trys it, someone who is a real file system whiz, not some hack, we won't really know just how reasonable it is.
That's the way computer science works; 1000 people say you can't do it, but one person does it, and it works, well, and suddenly everyone changes their opinion.
How many people, for example, really expected SGI to clean up XFS enough to merge into the official kernel?
I think it can be done. I can contemplate an algorithm that balances delete recovery priority policies with performance, and still comes out on top, with little speed degradations under most scenarios, and tolerable speed degradation in the worst case.
Between this imaginary file system I described and RAID, backups wouldn't be such a big deal.
A lot of people are backing up to hard disk anyway these days. What's the difference whether you are backing up to the same volume or a different one? You are just shifting where you need the space.
You'd still need some backups in case the filesystem fails or something, but It'd eliminate needing to use the backups very much.
I imagine something like rdiff-backup locally. You want to look at the filesystem the way it was 10 minutes ago? No problem. If you are going to take it this far, you might as well store deltas to updated files also.
It would have the advantage of being able to backup these snapshots too, rather than getting the nasty smearshot that you get when you backup a live filesystem these days.
I wish I knew more about filesystems... I might tackle this one day.
Slashdot Mirror
but at least we wouldn't have alien activists acting like idiots
:)
You haven't been on the Internet long, have you?
And not to go too far offtopic, but this is the reason why any human designed economy like socialism (currently called liberalism), is doomed to fail.
Capitalism has failings, but in general things work themselves out in the end. Monopolies collapse under their own weight. Abusive employers lose good employees which hurts them in the long run. Things like that.
I thought the whole premise of the hate crime laws was that the offender had to be doing something else illegal that was motivated by hate, and that was the only way it skirted the 1st amendment.
I guess all the people that say "slippery slope" is an invalid argument don't know much about governments or laws.
Normal antistatic bags have a very high resistance. Try it with a a pouch made of aluminum foil one day.
Of course it's going to be hard to read the display when it's in there.
If they had said "Kill all niggers" would you be so indignant?
Why would it make any difference, other than "Haitian" not being a derogatory term.
As someone with an Italian roots, I'd want Rockstar to defend their right to say "Kill all the fucking wop dagos!"
That was his point. The segregation isn't imposed by the majority, the segregation is imposed and enforced by the minorities themselves, because they don't really want to end racism and segregation either.
Because not everyone is getting software from the Internet. When the GPL was written, the Internet as we know it today didn't even exist.
The clause still makes sense in the Internet world though. Suppose I include a link to a third party site to get GPL code I distribute. They go out of business or change their domain name, or even take down the version I used because it became obselete. I'm now in violation of the GPL.
How is unix/linux behind Windows? The way I see it, it's way ahead of Windows in nearly all areas.
So I have a good uptime, until I try to upgrade a kernel, which requires me to dick with a lilo.conf and rebuild an initrc by hand. That's still downtime.
I don't have a problem with doing that, it's just unexpected, since Red Hat has had kernel upgrades be fully automatic for almost 5 years now.
There's a tradeoff between time spent fixing stability problems, and time spent doing mundane tasks that should be automatic. The mundane tasks are much more common.
Because there's a lot of untapped value there. When Debian is still using ghostscript 6, and thousands of other obselete packages, you have to sacrifice all the advances made in the last several years if you want to run Debian.
Testing/Sid isn't an option for production, since the Debian people won't commit to providing security fixes in any timely manner for those versions.
I think Red Hat used to have it right. A distro should be updated in a major way about once every year-18 months. Ideally you'd want to support the current version, and the immediate previous major version.
It worked great, I don't know why RH had to fuck it up.
They can release specifications without releasing code.
Well. Keep in mind that the Woody installer is several years old now.
That seems like a serious problem in itself to me!
Books and learning with people that already know are sufficient. The lecture model is obselete.
Well the danger with that, is the authors being deluged by people who just don't have the right dependancies installed.
Of course, it could be considered a bug if the configure script fails silently or ambiguously just because a dependancy was missing, but I see it a lot.
It'd really help if the authors would put very verbose missing dependancy messages, even as far as including a URL for the dependancies, if they aren't common.
One build that comes to mind that was total hell was Flightgear. That thing has dependancies all over the place, and not common things either.
didn't know what localhost referred to :P) but this is what you get without an affordable public education system...
That's a serious non sequitor. In my 12 years of "free" public education, I never took a class that taught anything about TCP/IP.
Most computer knowledge isn't taught, it's learned.
Heh, that's NOTHING.
My son fell down at an after school program, and broke his finger. We took him to the local walk-in clinic, and the doctor there took x-rays. She said she thought it should be referred to a specialist because the fracture was close to the joint.
She put a splint on it.
Took him to the specialist, they took x-rays, said it would be fine, and put a splint on it. Took maybe 5 minutes of the doctor's time, 5 minutes of the x-ray tech's time.
Get the bill, and the specialist charged my insurance company for $450 for "surgical services". That's not including the charge for the x-ray.
Called my insurance company (since it says right on the notice, to report suspected fraud). Insurance company says I need to call the specialist. Called the specialist and they said that's the normal way they bill "fracture care services".
So I guess there's nothing I can do. I had to pay 10% or something as a copayment, so it was like $40 for me, in addition to two other $20 copayments for the specialists x-ray and the walk-in clinic doctor.
Obviously I don't care about such small amounts of money, but frankly I'm stunned the insurance company would pay $450 for a fucking 5 minute glance at an xray, something the GP already did.
There's obviously corruption, bribes, kickbacks, all over the place in the medical insurance industry. If medical insurance is in "crisis", then it's because of this.
Just like the "energy crisis" with Enron, the "medical crisis" is a pile of shit. It's a coverup for the lies and corruption throughout the medical industry. In 5 or 10 years maybe we'll see some major scandal about medical billing. Until then the criminals behind this will continue to fleece everyone.
There's no excuse after all, Jeff Minter could probably be pryed away from his sheep and yaks long enough to score it for them. :)
If the people who were intelligent enough to not need university were not discriminated against, then a lot of people wouldn't waste time and money on university, developing unneeded debts.
The false idea that university education is somehow necessary for so many jobs is costing our society more than we realize.
where Britney Spears lives next door and cooks me pastries for breakfast on sundays!
She doesn't do that for you?
rdiff-backup and rsync-incremental backup shell scripts do something very similar, though it's only a smear-shot if the filesystem is changing.
We're using it to successfully do daily backups for the last 30 days for several TBs with very little overhead. No need for periodic full backups that would take days, since the diffs go back in time instead of forward.
XFS and ext3 I've both taken up to 50,000+ files in the same directory. There's no noticable performance lag on any of them. XFS does seem marginally better, but ext3 is about on-par. Once XFS is in the stock kernel for distros, I'll probably use it a lot more, it does seem to handle very large filesystems a little better.
I've always had trouble setting RAID stride for ext3 to actually see any performance increase, and the ext3 userland utils suck ass, with confusing documentation, and really bad defaults for large file systems. (The defaults don't scale)
I think Reiser's claims are overrated. It's nice to have a competitor, and I'm not anti-Reiser or anything, I just haven't seen any benefit, and I've run into serious bugs in Reiser in the past, even after all the Reiser fans were proclaiming stability.
And finally, the truth is, most people I know don't even delete files
I'm not discussing a file system for those people, I have storage and network file servers in mind.
Well, I hate defending MS, but it does sorta make sense to swap out stuff that hasn't been used in a while, even if there's lots of free RAM. Then if you start something that really needs tons of RAM, you won't have to wait for swap then, of stuff you weren't going to use anyway.
It's best if the algorithm has some idea what "big" is, but that's the difficulty in designing a VM system that works well with 128megs of ram or less and also 4+GB.
Gun ownership is a responsibility, not just a "basic human right."
Rights always require responsibility. Liberty requres plenty of responsibility, and yet you consider it a basic human right.
Our Constitution gives us this right
The Consistiution doesn't give rights. It merely enumerates what the founders thought were the most important inherent rights to form a long lasting stable state that offers freedom.
Note that later in the bill of rights they mention "rights retained by the people". This should be clear enough evidence that the bill of rights should not be construed to be granting rights, only to enumerate rights that already exist.
for one explicit purpose -- to maintain a militia
This is a dead horse debate over wording, but the way the 2nd is written, it's clearly not meant to be the sole purpose. If I say "Money is good, I should go get a job", that doesn't mean I want to get a job solely because "money is good", it's simply one reason that I found compelling.
In other words, owning a firearm is part of every man's civic duty to protect his country and resist oppresion and is not an inherent right to go waving about like you automatically deserve to handle a dangerous weapon simply for breathing.
I think you misunderstand the term "right". I think you've internalized the liberal definition of "human rights", which is definitely streched from what the founders intended. They made it clear that rights require a certain level of responsibility in exercising of the rights. They didn't have a problem with punishing people who abused thier rights.
The supreme court has been generally been very careful to avoid "prior restraint" in 1st amendment issues. The idea is that everyone should have the ability to exercise their rights, and they shouldn't be punished until and unless they abuse their rights.
As far as whether the line is arbitrary or not, you mention cost/benefit ratios. That's not how constitutional law has ever been interpreted. There's obviously a high cost in social stability in allowing the KKK to march, and very little social benefit. Cost/benefit has never been a factor when it comes to fundamental rights. If it ever becomes a major factor, we are in serious trouble.
In essence, the 2nd Amendment is unfortunately obsolete.
Well, the founders couldn't have possibly forseen the Internet, so the 1st amendment must be obselete too. Hell, why don't we repeal all 10 amendments in the bill of rights, they are pretty much all "obselete" under your argument.
The 2nd amendment doesn't say "as long as it's easy", or "until better weapons are invented". Your argument flies in the face of the Bill of Rights as a whole.
I think a balance can be found.
Until someone trys it, someone who is a real file system whiz, not some hack, we won't really know just how reasonable it is.
That's the way computer science works; 1000 people say you can't do it, but one person does it, and it works, well, and suddenly everyone changes their opinion.
How many people, for example, really expected SGI to clean up XFS enough to merge into the official kernel?
I think it can be done. I can contemplate an algorithm that balances delete recovery priority policies with performance, and still comes out on top, with little speed degradations under most scenarios, and tolerable speed degradation in the worst case.
Between this imaginary file system I described and RAID, backups wouldn't be such a big deal.
A lot of people are backing up to hard disk anyway these days. What's the difference whether you are backing up to the same volume or a different one? You are just shifting where you need the space.
You'd still need some backups in case the filesystem fails or something, but It'd eliminate needing to use the backups very much.
I imagine something like rdiff-backup locally. You want to look at the filesystem the way it was 10 minutes ago? No problem. If you are going to take it this far, you might as well store deltas to updated files also.
It would have the advantage of being able to backup these snapshots too, rather than getting the nasty smearshot that you get when you backup a live filesystem these days.
I wish I knew more about filesystems... I might tackle this one day.
Well, it uses a good chunk of RAM for cache. Windows VM never was highly acclaimed or anything.