Depends. If they make an effort to at least pretend to some sort of neutrality, and routinely acknowledge facts that are embarrasing to the powers that be (while, of course, subtly 'spinning' those facts so as to be less damaging) that would be like CNN. If they make no effort to such pretences, and tend to stick their fingers in their ears and start chanting, screaming, and mudslinging whenever someone tries to mention the embarrasing facts, that'd be Faux News.
Uhh, no. Windows is a GUI shell. All recent version run on the NT kernel, older ones often ran on a DOS kernel. Don't let the marketing bundle fool you. The kernel and the GUI shell are still two quite different things, even if they're provided by a company that only sells them together on the same media and tries to obscure the difference wherever possible.
The HURD is one of many possible kernels on which the GNU system can run. The first GNU kernel was called TRIX, and was developed at MIT in the 80s. It was never very advanced, and much work was done instead using the GNU system on proprietary kernels from folks like Sun and HP as a result of its limitations. The FSF having very limited resources had to prioritise and economise. Rather than pull scarce resources off other, equally crucial projects to bring TRIX up to snuff, it was decided to pursue rights from CMU to use the Mach kernel. This would jumpstart the kernel project for free, without diverting resources needed elsewhere.
The HURD did, actually, exist in 1991. In that year, after several years of negotiation with CMU, all necessary issues were resolved and the longstanding plan to build a new GNU kernel on top of Mach began. Linus famously remarked in an early usenet posting that he expected in a few years everyone would be using it and his project wouldn't really matter.
What happened instead, of course, is that he was so successful at producing a good, working free kernel, people started running GNU on top of it, and the HURD team felt free to pursue long-term experimental strategies instead of feeling pressure to produce a working product quickly.
You can run the HURD, although it's still experimental and not recommended for production systems. The most popular way to set it up is through debian.
No, the GNU system was a work in progress already long before linux started. It was an OS project from day one.
Linux was made possible by it, and the linux kernel made it possible for GNU to exist in full freedom in turn.
There were plenty of GNU systems before linux, they were just hybrids, built on top of proprietary kernels. This is called bootstrapping. There's no other way to do it.
I think there current approach, as opposed to the 'splash' strategy you suggest, is better both for them and us.
The last thing I want to see is any company with anything like a monopoly on linux.
And as to them not caring what OS their customers use, you say that like it's a bad thing?
I see it as good all the way around. I'd be no more interested in some kind of linux lockin than I am in windows lockin. IBM currently seems to be hewing to a real old-fashioned business principle called 'the customer is king.' That's how it should be. I wish more companies got in that mode.
Not at all. They don't owe me a thing, and I don't owe them a thing. That's because I don't buy their hardware. I *would* be happy to buy their hardware, for myself, for friends, for clients, etc. if they would learn to work with the community, but it's entirely up to them.
But they are damn sure *attempting* to restrict my freedom when they *attempt* to sell that hardware to me. Because that's what that hardware does. It will only work properly if the user gives up their freedom and their control of their machine. How much clearer can it be?
And you're still completely missing the point. It's not a *risk* of something bad that may or may not happen, it's a _certainty_ - the moment you allow an opaque binary to run inside the kernel on your machine, you have given up control of that machine. Period. You no longer have even theoretical control of the machine, they own it. Whether or not they, by choice or otherwise, proceed to do other bad things, that one's done and you no longer have any say in the matter. All you can do is hope they won't screw you any harder.
Of course it can happen. Another poster remembered a specific example - the drivers for the old pre-ata drives that shipped with the PC/XT were recently dropped. They never shipped on a machine with a 386 processor best I know, but some did wind up in 386s anyway. There might be a few left in use, but it seems doubtful.
Proprietary drivers often go unsupported *much* quicker.
Immature projects aren't a problem. They're the grist for the evolutionary mill. The largest number will always fail. The best of the lot will progress. That's just how it works. If there's a problem it's your expectation of somehow being provided with wheat without some chaff being generated in the process.
If you need mind mapping software, for instance, don't run grab a half-finished java hack, tear it up, then complain you can't get your work done with it. It makes you sound like an astroturfer. Particularly when this is used as evidence that 'linux' isn't ready and the app you're basing this judgement on isn't even a linux app!
Try Kdissert, if you really need a mind-mapping tool.
You're missing the point. Even if it does nothing *else* malicious, a binary kernel driver *still* 'takes over your computer' the moment it's installed. It's got full access, it can do exactly what it's developer coded (whether that's what he intended or not) and you have NOTHING to say about the matter. You can't examine it, you can't modify it, your computer has been 'pwned.' Now, if you're lucky, maybe it won't go on to screw you in other ways, but it's still in charge of your computer, and you aren't anymore.
There are a half-dozen mom and pops computer shops in my area, and we're talking out in the boonies.
Sadly, only one of them is conversant with linux, but he's also the one with an increasing, rather than a decreasing, customer base.
For a business this should be an obvious thing. Don't they teach you to avoid getting locked into a monopoly supplier in business school anymore? If you need something fixed, changed, expanded in windows, there's really only one entity with the ability to do that, and they're much too big to care about your desires unless you're in the fortune 500, at least.
Build your infrastructure around *nix and you've got so many options for support and supply it makes the head swim. That means they compete, and you win.
The point is, look at what you pay for the software. Not just in monetary terms.
Nvidia doesn't need to put a backdoor in their driver for the cost of it to be too high, because the known cost, without that, is still the users freedom. Their freedom to study how their system works. Their freedom to change how it works, or hire someone to change it for them. The freedom to run WHATEVER OS on it you choose. Sure, they're releasing linux drivers, for now. How's that help you if you want to run BSD? Or Plan9, or BeOS, or anything else? It doesn't. It may not even work when the next kernel comes out.
At the most basic level, it takes away the customers ability to control the hardware they've bought and paid for, even if it doesn't have any unwanted features.
There are plenty of practical problems that go along with that, statistically speaking. More bugs, yes, but more importantly a helplessness against the bugs. If your video driver is buggy and crashing your system, or worse, there are many people out there with the expertise to help debug it - but if that driver is blobware they can't help you. You're reduced to complete dependence on the vendor - who probably doesn't even think of you as a customer. Their customers are other big companies - you are a commodity to them. If you don't want to be that, you have to insist on keeping your freedom.
Now, as to what you were talking about, of course bugs and malware can be inserted in free code - but not nearly as easily, and of course bugs and malware can be detect in unfree blobs - but not near as easily. If that's your only concern, you're an 'open source' person, and that's fine, you still don't want blobware.
But the issues here are much deeper than the practical - the philosophical is much more important, the practicalities are ultimately reflections of the philosophies we live by, consciously or unconsciously. If you don't mind being a commodity that big corporations buy and sell - an 'eyeball' to the media companies and advertisers, for instance, rather than a customer, then I guess you won't mind having no control of the computer hardware you use either. You'll be happy with the blobware running your computer on behalf of its maker, and all their real customers that they sell you to. It'll get you clippy, and hassle-free hollywood movies, and endless britney spears videos, so why should you care if it means your computer really belongs to MS and is for sale to the highest bidders?
That's the issue here, at the core. Everything else follows from it, even the practicalities, because they're a simple consequence of the fact that freedom works. But even if it didn't work so well, some of us would still insist on keeping it.
I think you largely hit the nail on the head there.
I'm an example. I'm just a computer user, not a kernel hacker or anything like that. I've been using linux for well over a decade, and haven't used a windows machine in around half that time. I'm always amused at the 'is linux ready for the desktop' crap that keeps getting recycled every 6 months. Of course it's ready. It's been ready for a decade. The question is 'are you ready?'
Yes, there's some things that toy computers do that linux won't. So what. That's the stuff I used to spend half my time finding ways to turn off when I was on windows anyway.
Hardware support could definitely be better, but then again, the only solution to that is long term. Don't buy hardware that doesn't work.
You know, it occurs to me that this may be one of the reasons some hardware manufacturers resist supporting linux. With Windows and secret interfaces, they get to cut driver support with a fair expectation of forcing people to buy new hardware.
So by your logic, if I give you a binary blob that 'gets your stuff done' that's it, end of story, right?
Now what if that blob also lets me take over your computer and do whatever I want with it? Maybe I'll use it to send a few million spams... or maybe I'll just snoop through your private documents and check out your pr0n collection. Either way, doesn't matter, since you 'got your stuff done' right?
What you're doing is looking at one side of the equation - benefit - but not at the other - cost.
What 'stuff' exactly do you do with your computers that's worth more than your freedom?
Drivers for fully supported hardware are kept right in there in the kernel tree on kernel.org.
Whenever the internal kernel interfaces change (which is really pretty rarely) kernel programmers also check all those drivers and make any changes necessary.
Once a device is supported, it's very nearly perpetual. It's rare for drivers to be removed, and usually when they are it's because they've been superceded (the hardware still works, the support is just being done more elegantly, for instance when 2.6.16 was released amdtp and cmp had been removed, but that was because the hardware they supported is now supported with libiec61883.)
For a device that was once supported to actually be dropped, there has to be a major kernel change combined with no one in the kernel development community (paid or volunteer) having the motivation to update it. This means, no paying customers of RedHat, SuSE, etc. using it, no kernel hackers have one in a still functional system at home, etc. And even then, you're free to grab the source for the last working version and update it yourself, or pay someone else to do it.
Okay, this weeks drivers might not work with next weeks kernel, but this is a problem with the linux kernel not having the same backwards compatibility as windows. Can hardly blaim nvidia for that.
The fact that the kernel isn't obligated to accrete cruft continually to preserve backward compatibility is an advantage, not a problem.
Nvidias unwillingness to simply document their product so it can be properly supported is not something you can blame on anyone but them. Old-think dies hard, but it does die.
I don't think the problem is Linspire using non-free software. Lots of folks do that.
The problem was the tone he was taking, talking this up as a wonderful advance that everyone should emulate, lionizing the supposed 'freedom' to 'choose' to be unfree. That's what really set PJ off. And it left a bad taste in my mouth when I read it too.
Keep in mind that when WinXP came out I finally said enough is enough and quit using Windows completely. I've never touched any windows version later than Win2k...
OK... what's the correct procedure to clean up the registry when an errant program makes a mess of it?
Restore the backup registry. After making note of all the differences, so you can try to manually restore and fix the necessary entries for making the POS that caused the problem work, if that's actually required.
Honestly, in many cases it's less work just to wipe the machine and reinstall though. Registry is annoying.
Or... what options are there to make Windows more flexible in a heterogenous environment without using SFU?
Cygwin? Not sure what you're getting at here exactly. FTP, HTTP, telnet/SSH, SMTP, all can be made to work more or less correctly on Windows as I recall. I always tried to avoid using Windows-specific things on the network, and the places where I was ordered to do that weren't heterogenous...
Or... What is the single most important change you can make to your Windows installation to prevent or reduce the occurrence of spyware on your system (hint: it's not spyware removal tools)?
Lock down, remove, or (wherever possible) prevent the initial installation of IE and Outlook. LitePC has tools that I used to use to great effect here, some are free.
Of course to be truly secure, you need to yank all removable media, make sure the box is never plugged into a network and keep strict physical control.;)
Unless someone has the right answers to those questions I don't think they can say they know much about computers.
I can't agree, since there are a few that have been fortunate enough to never have to work with windows and never needed to learn the ugly workarounds it requires. Doesn't mean they don't know computers...
Slashdot Mirror
Are you trolling or just incredibly stupid?
Depends. If they make an effort to at least pretend to some sort of neutrality, and routinely acknowledge facts that are embarrasing to the powers that be (while, of course, subtly 'spinning' those facts so as to be less damaging) that would be like CNN. If they make no effort to such pretences, and tend to stick their fingers in their ears and start chanting, screaming, and mudslinging whenever someone tries to mention the embarrasing facts, that'd be Faux News.
Exactly. The sooner projects move to v3 or later, the better.
Uhh, no. Windows is a GUI shell. All recent version run on the NT kernel, older ones often ran on a DOS kernel. Don't let the marketing bundle fool you. The kernel and the GUI shell are still two quite different things, even if they're provided by a company that only sells them together on the same media and tries to obscure the difference wherever possible.
This is exactly why it's so important to move to the GPLv3.
The HURD is one of many possible kernels on which the GNU system can run. The first GNU kernel was called TRIX, and was developed at MIT in the 80s. It was never very advanced, and much work was done instead using the GNU system on proprietary kernels from folks like Sun and HP as a result of its limitations. The FSF having very limited resources had to prioritise and economise. Rather than pull scarce resources off other, equally crucial projects to bring TRIX up to snuff, it was decided to pursue rights from CMU to use the Mach kernel. This would jumpstart the kernel project for free, without diverting resources needed elsewhere.
The HURD did, actually, exist in 1991. In that year, after several years of negotiation with CMU, all necessary issues were resolved and the longstanding plan to build a new GNU kernel on top of Mach began. Linus famously remarked in an early usenet posting that he expected in a few years everyone would be using it and his project wouldn't really matter.
What happened instead, of course, is that he was so successful at producing a good, working free kernel, people started running GNU on top of it, and the HURD team felt free to pursue long-term experimental strategies instead of feeling pressure to produce a working product quickly.
You can run the HURD, although it's still experimental and not recommended for production systems. The most popular way to set it up is through debian.
No, the GNU system was a work in progress already long before linux started. It was an OS project from day one.
Linux was made possible by it, and the linux kernel made it possible for GNU to exist in full freedom in turn.
There were plenty of GNU systems before linux, they were just hybrids, built on top of proprietary kernels. This is called bootstrapping. There's no other way to do it.
I think there current approach, as opposed to the 'splash' strategy you suggest, is better both for them and us.
The last thing I want to see is any company with anything like a monopoly on linux.
And as to them not caring what OS their customers use, you say that like it's a bad thing?
I see it as good all the way around. I'd be no more interested in some kind of linux lockin than I am in windows lockin. IBM currently seems to be hewing to a real old-fashioned business principle called 'the customer is king.' That's how it should be. I wish more companies got in that mode.
Not at all. They don't owe me a thing, and I don't owe them a thing. That's because I don't buy their hardware. I *would* be happy to buy their hardware, for myself, for friends, for clients, etc. if they would learn to work with the community, but it's entirely up to them.
But they are damn sure *attempting* to restrict my freedom when they *attempt* to sell that hardware to me. Because that's what that hardware does. It will only work properly if the user gives up their freedom and their control of their machine. How much clearer can it be?
And you're still completely missing the point. It's not a *risk* of something bad that may or may not happen, it's a _certainty_ - the moment you allow an opaque binary to run inside the kernel on your machine, you have given up control of that machine. Period. You no longer have even theoretical control of the machine, they own it. Whether or not they, by choice or otherwise, proceed to do other bad things, that one's done and you no longer have any say in the matter. All you can do is hope they won't screw you any harder.
I think so.
Of course it can happen. Another poster remembered a specific example - the drivers for the old pre-ata drives that shipped with the PC/XT were recently dropped. They never shipped on a machine with a 386 processor best I know, but some did wind up in 386s anyway. There might be a few left in use, but it seems doubtful.
Proprietary drivers often go unsupported *much* quicker.
It can't. But the proven part can act as an overseer of the unproven part. That's what it means to be the kernel.
Immature projects aren't a problem. They're the grist for the evolutionary mill. The largest number will always fail. The best of the lot will progress. That's just how it works. If there's a problem it's your expectation of somehow being provided with wheat without some chaff being generated in the process.
If you need mind mapping software, for instance, don't run grab a half-finished java hack, tear it up, then complain you can't get your work done with it. It makes you sound like an astroturfer. Particularly when this is used as evidence that 'linux' isn't ready and the app you're basing this judgement on isn't even a linux app!
Try Kdissert, if you really need a mind-mapping tool.
You're missing the point. Even if it does nothing *else* malicious, a binary kernel driver *still* 'takes over your computer' the moment it's installed. It's got full access, it can do exactly what it's developer coded (whether that's what he intended or not) and you have NOTHING to say about the matter. You can't examine it, you can't modify it, your computer has been 'pwned.' Now, if you're lucky, maybe it won't go on to screw you in other ways, but it's still in charge of your computer, and you aren't anymore.
There are a half-dozen mom and pops computer shops in my area, and we're talking out in the boonies.
Sadly, only one of them is conversant with linux, but he's also the one with an increasing, rather than a decreasing, customer base.
For a business this should be an obvious thing. Don't they teach you to avoid getting locked into a monopoly supplier in business school anymore? If you need something fixed, changed, expanded in windows, there's really only one entity with the ability to do that, and they're much too big to care about your desires unless you're in the fortune 500, at least.
Build your infrastructure around *nix and you've got so many options for support and supply it makes the head swim. That means they compete, and you win.
Well, so far as the articles in so-called news outlets, they seem to rehash them about every 6 months.
You're completely missing the point.
The point is, look at what you pay for the software. Not just in monetary terms.
Nvidia doesn't need to put a backdoor in their driver for the cost of it to be too high, because the known cost, without that, is still the users freedom. Their freedom to study how their system works. Their freedom to change how it works, or hire someone to change it for them. The freedom to run WHATEVER OS on it you choose. Sure, they're releasing linux drivers, for now. How's that help you if you want to run BSD? Or Plan9, or BeOS, or anything else? It doesn't. It may not even work when the next kernel comes out.
At the most basic level, it takes away the customers ability to control the hardware they've bought and paid for, even if it doesn't have any unwanted features.
There are plenty of practical problems that go along with that, statistically speaking. More bugs, yes, but more importantly a helplessness against the bugs. If your video driver is buggy and crashing your system, or worse, there are many people out there with the expertise to help debug it - but if that driver is blobware they can't help you. You're reduced to complete dependence on the vendor - who probably doesn't even think of you as a customer. Their customers are other big companies - you are a commodity to them. If you don't want to be that, you have to insist on keeping your freedom.
Now, as to what you were talking about, of course bugs and malware can be inserted in free code - but not nearly as easily, and of course bugs and malware can be detect in unfree blobs - but not near as easily. If that's your only concern, you're an 'open source' person, and that's fine, you still don't want blobware.
But the issues here are much deeper than the practical - the philosophical is much more important, the practicalities are ultimately reflections of the philosophies we live by, consciously or unconsciously. If you don't mind being a commodity that big corporations buy and sell - an 'eyeball' to the media companies and advertisers, for instance, rather than a customer, then I guess you won't mind having no control of the computer hardware you use either. You'll be happy with the blobware running your computer on behalf of its maker, and all their real customers that they sell you to. It'll get you clippy, and hassle-free hollywood movies, and endless britney spears videos, so why should you care if it means your computer really belongs to MS and is for sale to the highest bidders?
That's the issue here, at the core. Everything else follows from it, even the practicalities, because they're a simple consequence of the fact that freedom works. But even if it didn't work so well, some of us would still insist on keeping it.
I think you largely hit the nail on the head there.
I'm an example. I'm just a computer user, not a kernel hacker or anything like that. I've been using linux for well over a decade, and haven't used a windows machine in around half that time. I'm always amused at the 'is linux ready for the desktop' crap that keeps getting recycled every 6 months. Of course it's ready. It's been ready for a decade. The question is 'are you ready?'
Yes, there's some things that toy computers do that linux won't. So what. That's the stuff I used to spend half my time finding ways to turn off when I was on windows anyway.
Hardware support could definitely be better, but then again, the only solution to that is long term. Don't buy hardware that doesn't work.
Looks like an interesting thread. Sadly their server is smoking already. I'll check it out in the morning.
You know, it occurs to me that this may be one of the reasons some hardware manufacturers resist supporting linux. With Windows and secret interfaces, they get to cut driver support with a fair expectation of forcing people to buy new hardware.
So by your logic, if I give you a binary blob that 'gets your stuff done' that's it, end of story, right?
Now what if that blob also lets me take over your computer and do whatever I want with it? Maybe I'll use it to send a few million spams... or maybe I'll just snoop through your private documents and check out your pr0n collection. Either way, doesn't matter, since you 'got your stuff done' right?
What you're doing is looking at one side of the equation - benefit - but not at the other - cost.
What 'stuff' exactly do you do with your computers that's worth more than your freedom?
Drivers for fully supported hardware are kept right in there in the kernel tree on kernel.org.
Whenever the internal kernel interfaces change (which is really pretty rarely) kernel programmers also check all those drivers and make any changes necessary.
Once a device is supported, it's very nearly perpetual. It's rare for drivers to be removed, and usually when they are it's because they've been superceded (the hardware still works, the support is just being done more elegantly, for instance when 2.6.16 was released amdtp and cmp had been removed, but that was because the hardware they supported is now supported with libiec61883.)
For a device that was once supported to actually be dropped, there has to be a major kernel change combined with no one in the kernel development community (paid or volunteer) having the motivation to update it. This means, no paying customers of RedHat, SuSE, etc. using it, no kernel hackers have one in a still functional system at home, etc. And even then, you're free to grab the source for the last working version and update it yourself, or pay someone else to do it.
Try to do that with a binary driver.
The fact that the kernel isn't obligated to accrete cruft continually to preserve backward compatibility is an advantage, not a problem.
Nvidias unwillingness to simply document their product so it can be properly supported is not something you can blame on anyone but them. Old-think dies hard, but it does die.
I don't think the problem is Linspire using non-free software. Lots of folks do that.
The problem was the tone he was taking, talking this up as a wonderful advance that everyone should emulate, lionizing the supposed 'freedom' to 'choose' to be unfree. That's what really set PJ off. And it left a bad taste in my mouth when I read it too.
Uhh yes, actually you could. It was called MSX-Dos. Microsoft put it out for a few years in the early 80s.
Keep in mind that when WinXP came out I finally said enough is enough and quit using Windows completely. I've never touched any windows version later than Win2k...
Restore the backup registry. After making note of all the differences, so you can try to manually restore and fix the necessary entries for making the POS that caused the problem work, if that's actually required.
Honestly, in many cases it's less work just to wipe the machine and reinstall though. Registry is annoying.
Cygwin? Not sure what you're getting at here exactly. FTP, HTTP, telnet/SSH, SMTP, all can be made to work more or less correctly on Windows as I recall. I always tried to avoid using Windows-specific things on the network, and the places where I was ordered to do that weren't heterogenous...
Lock down, remove, or (wherever possible) prevent the initial installation of IE and Outlook. LitePC has tools that I used to use to great effect here, some are free.
Of course to be truly secure, you need to yank all removable media, make sure the box is never plugged into a network and keep strict physical control. ;)
I can't agree, since there are a few that have been fortunate enough to never have to work with windows and never needed to learn the ugly workarounds it requires. Doesn't mean they don't know computers...
So anyhow, you tell me. How'd I do?