This is one of the telltale remaints of the BSD-derived TCP/IP stack that NT/XP uses.
Even the page you link to disproves this (incorrect as usual when it comes to Slashdot and Windows) meme. NT hasn't used a BSD-derived TCP/IP stack since NT 3.5, and Windows 9x has never used it.
In unix, security and permissions are the foundation, on top of which everything is built. In windows, security is a hack that was added on later with no due consideration during the initial design phase of windows.
Only because people always need admin priveleges at the most inopportune times. Sure, you only need to be Admin for 2 seconds, but if you're doing anything technical with the system, you need them every 5-10 minutes. In these situations Run As is at best, cumbersome, and in many cases outright incapable.
What on Earth are you doing that needs Administrator privileges every "5 - 10 minutes" ?
And if you really *do* need it that often, why aren't you just keeping an Administrator-level cmd or explorer window open from which you can launch things at will ?
If Windows wanted to be truly innovative there would be a way to supply an Admin password, temporarily upgrade privs to Admin, and then have a button available to immediately downgrade.
Which is, uh, basically what Run As does.
I'd like the IP on that idea since it'll be central to making a GUI OS both secure and functional.
Ignoring that there's massive amounts of prior art, it won't have much effect on security, because people would just run as Admin all the time anyway and never hit the button to "downgrade" their rights again.
With Apple now shipping x86 computers people are starting to realize that yea Apple hardware really is higher than average PC quality.
No, it's not. They use the same CPUs, the same hard disks, the same memory, the same chipsets, etc. In many cases they're even built on the same assembly lines.
The "Macs have better hardware" has been nothing more than Mac Zealotry since *at least* the PCI PowerMacs, and probably longer (particularly when you take into account the higher prices).
Apple x86 machines are jumping to the top of the list for performance vs price. Something that Apple Fanatics have been saying for years but no one really believed them.
No, they're still _very_ average in terms of price/performance, *unless* you attach significant value to the form factor. If you're happy with a beige box, it's not difficult to better the price/performance of a Mac.
The value in a Mac is not - and never has been - the hardware. It's the software that is packaged along with it.
Sudo only delivers real benefits if it is properly setup. That is, an/etc/sudoers that restricts which users (or groups) can run which commands, on which machines, etc, etc.
If your/etc/sudoers is only a few lines long and basically does (and is used for) little more than allow users X, Y and Z get a root shell, then there's really little point to using it at all and you may as well just save the annoyance of having to run 'sudo bash', 'sudo su', 'sudo -i' - or whatever other incantation you use to get a root shell - and let people use root all the time.
I do this on all of my servers. And even more, there's this great command if you really need a root shell:
sudo su
Congratulations. You have now completely removed almost every reason for using sudo in the first place.
If all you use sudo for is starting a root shell once you've logged in, then save yourself the hassle and just login as root, because you're circumventing basically every benefit sudo offers.
Programs should not access to "per-user settings", they should only have access to Per-User/Per Application. An Application should only be able to change it's own settings, not the setting of an entire user's registry space.
This is not really a problem with, or criticism of, the Registry, but one of the whole user-based security model (and one valid for rall major platforms, at least as they are typically used).
You could facilitate what you want by running each application in its own user account. A bit messy, but effective.
This way, removing registry entries would be easier.
Not easier, safer. The operation would be identical, there would be just less (if any) opportunity for bugs in the uninstallation process to modify other data.
Look, it's not worth arguing that the registry is a good design... we wouldn't be having this conversation if there was no problem.
I have never said the Registry was flawless. The basic design, however, is sound. It's certainly miles ahead of anything based around editing text files.
But I hope MS is breaking compatibility on this little gem in Vista, because if they don't, I'm not sure Vista will offer the ability to fend off badly written (and malicious)programs from hijacking the registry easily, just as they do in XP/2000/NT.
Programs could hijack textfiles jsut as easilly (more so, if anything).
When used properly, the registry really isn't all that bad. While I think.ini files were better, the registry isn't necessarily a bad thing. However, the security sucks.
The security in the Registry is excellent. Each individual key has per-user ACLs. It's much better security than plain text files are, or could practically be.
The *problem* - as you say later - is that lazy/incompetent/ignorant application developers write their applications badly.
This should be locked down just like the rest of Program Files, so admin/root is the only one who can make changes. Put per-user settings in Documents and Settings/username/Application Data.
This is exactly how the Registry works - system-wide setting go in HKEY_LOCAL_MACHINE and per-user settings go in HKEY_USERS\{SID}. Again, the problem is developers, not the Registry.
The only real problem with the Registry is that it is completely backed onto binary data files. Of course, when you consider it was specified and designed back in about 1990 when a 33Mhz 386 was a blazingly fast machine, that's an understandable design choice.
I'd love to help, but I can't stop laughing from seeing the words "elegant" and "windows" together. And, is it just me, or is the summary itself dripping with sarcasm?
It's not just you - the whole thing pretty much screams out "troll".
Yeah, but "abusing monopoly position", what the hell does that mean? That is so vauge as to be meaningless. A law like that isn't a law, it is a catch-all disclaimer to allow the government to go after any company it feels like.
It's not just the "abuse" part that's vague, either. "Monopoly" is in no way formally and strictly defined, and a company really has no way of knowing that it is a monopoly until a court of law finds it to be one (at which point completely normal business practices suddenly become "monopoly abuse"),
Uh, Linux and OS X didn't need to be patched to fix RPC exploit worms that rebooted two-thirds of the world's computers, or database server worms that gave SQL Server the dubious distinction of being the platform for the fastest spreading worm in history.
Mainly because Linux and OS X - combined - still only represent around 1 in every 20-odd computers.
Even a "perfect storm" of Linux or OS X worms wouldn't have anything close to the impact of a relatively low-key Windows worm. There's simply not enough machines running either of them out there for it to spread especially fast or hit especially hard.
There have been *serious* flaws in OS X and Linux that have been patched since their release (more for Linux, obviously, since it's been around longer). Don't kid yourself otherwise. The reason these flaws have little noticable impact is because, relatively speaking, OS X and Linux have next to zero internet presence.
Try dropping a completely unpatched, default install of Redhat 7.x, 8.x or 9.x on a fast internet connection and see how long it lasts.
Just like legalese can be used to frustrate or to clarify in a way no other mode of language can.
Legalese is required because the people writing it know they can't trust dishonest people (like them) to do the right thing - thus what they can and can't do needs to be spelt out in painfully explicit detail.
Suppose the poster is a business major who has been thrust into the IT/S division of his company, asking us business folk if he should have to learn these ridiculous technical terms in order to communicate with the people he has to deal with every day.
Difference is technical terms have real, concrete, well-defined meanings and are used to *clarify* discussion, whereas corporate-speak is typically the exact opposite.
Eh. NTFS's performance still goes to hell when it fragments heavily, which happens rapidly when the drive is near full. This sucks if you're moving around large amounts of data. Windows gets very, very unusable (and very unstable) when the filesystem starts to get seriously fragmented. Yah, you shouldn't fill NTFS volumes over 85% full - but it's not like the operating system even warns you that all hell is about to break loose.
I've yet to experience this sort of behaviour on any Windows system, although I've heard it bandied around frequently. I suspect it's much like that "Internet Explorer is part of the Windows kernel" that gets similarly parroted on Slashdot. Sure you're not repeating hearsay dating from NT 3.x and 4.0 ?
So "Windows, on well made hardware, with good drivers, politely used, is very stable"?
Yes. I can't even remember the time I saw a Windows NT machine crash where faulty hardware or bad drivers wasn't directly responsible (and easily identifiable as such).
[...]as long as you don't connect to the internet, [...]
Bollocks.
[...] or run any apps on it, [...]
More bollocks.
or use any devices (even well made ones) that use the same portion of the kernelspace at once.
Just like any other OS (assuming you're trying to say what I think you're trying to say).
The difference is that Windows doesn't do a good job of implementing levels of trust.
You don't appear to have the requisite technical knowledge to make that judgement.
What I don't expect is that they'll be able to bring down my OS, or any portion of it, when they do.
If you've never seen Linux or OS X kernel panic, you haven't been using either for very long.
Further, I expect that drivers written by Microsoft or that are Microsoft certified shouldn't have any compatibility problems. Otherwise, why did they get certified?
The USB subsystem shouldn't have the privileges to crash the rest of the OS. That's a design flaw.
Uh, it's a *hardware driver*. Hardware drivers (mostly) run in kernel space. Hardware drivers can most certainly crash the system.
This is not a design flaw, it's a design choice - one made for performance. Rest assured that faulty drivers are just as capable of bringing down Linux and OS X, as they are Windows.
This is just a single example that is typical of what you see in Windows, and the logic behind why I say that Windows itself is not stable. There are many, many others.
It's not typical behaviour at all. Your hardware is broken, and this is not Windows's fault.
Linux runs on tons more platforms and configurations than Windows does, and it has never had a problem with stability as far as I'm aware.
There are no shortage of stability problems with Linux when you get into the crappier end of the hardware scale, just like Windows (as I have been reminded over the last week losing more hair to machines with VIA chipsets).
MS isn't stable, because they don't care about stability
Windows on well made hardware, and with good drivers, is extremely stable. If your Windows installation is not stable, Windows is not the problem - it's either broken hardware or bad drivers.
The system design and development model has led to two things, a shortage of privilage escalation flaws (flaws isn't good enough, they have to allow a user account to gain root under conditions the virus can create) and a short lifespan of any such flaws that exist.
Very few pieces of malware need elevated privileges to do their "job".
grandchild.jpg.exe can never work on linux, period. You have to get the user to open a prompt cd to/home/granny/.email/files then chmod +X grandchild.jpg.exe, THEN./grandchild.jpg.exe (in linux you have to create a launcher to execute a file in the gui, double-clicking will not work.
Users are happy to open password protected zipfiles to get at the empty promise of porn, viagra and free iPods, and you think a simple command like "chmod u+x file" is going to slow them down ?
Windows users are prepared for viruses and the reason Linux users do not sweat them much is not because linux viruses do not exist; it is because system design makes their impact minimal.
>More people spouting the FUD about how the GPL is "viral".
It's not FUD. It might not be the nicest way of putting it, but "viral" describes what the GPL does and was designed to do. Some GPLed code getting into a proprietry, commercial software codebase would have the potential to be very bad, especially if it happened deliberately.
Slashdot Mirror
Even the page you link to disproves this (incorrect as usual when it comes to Slashdot and Windows) meme. NT hasn't used a BSD-derived TCP/IP stack since NT 3.5, and Windows 9x has never used it.
False on both counts.
What on Earth are you doing that needs Administrator privileges every "5 - 10 minutes" ?
And if you really *do* need it that often, why aren't you just keeping an Administrator-level cmd or explorer window open from which you can launch things at will ?
If Windows wanted to be truly innovative there would be a way to supply an Admin password, temporarily upgrade privs to Admin, and then have a button available to immediately downgrade.
Which is, uh, basically what Run As does.
I'd like the IP on that idea since it'll be central to making a GUI OS both secure and functional.
Ignoring that there's massive amounts of prior art, it won't have much effect on security, because people would just run as Admin all the time anyway and never hit the button to "downgrade" their rights again.
It can, and does.
These people aren't even close to a typical user.
If this trend is to show what the future holds, Apple might in the end be sorry about what they've done to their business.
Their business is selling Macs. Someone running Windows on a Mac has no negative impact on this at all, as Apple has still sold them a Mac.
No, it's not. They use the same CPUs, the same hard disks, the same memory, the same chipsets, etc. In many cases they're even built on the same assembly lines.
The "Macs have better hardware" has been nothing more than Mac Zealotry since *at least* the PCI PowerMacs, and probably longer (particularly when you take into account the higher prices).
Apple x86 machines are jumping to the top of the list for performance vs price. Something that Apple Fanatics have been saying for years but no one really believed them.
No, they're still _very_ average in terms of price/performance, *unless* you attach significant value to the form factor. If you're happy with a beige box, it's not difficult to better the price/performance of a Mac.
The value in a Mac is not - and never has been - the hardware. It's the software that is packaged along with it.
If your /etc/sudoers is only a few lines long and basically does (and is used for) little more than allow users X, Y and Z get a root shell, then there's really little point to using it at all and you may as well just save the annoyance of having to run 'sudo bash', 'sudo su', 'sudo -i' - or whatever other incantation you use to get a root shell - and let people use root all the time.
Congratulations. You have now completely removed almost every reason for using sudo in the first place.
If all you use sudo for is starting a root shell once you've logged in, then save yourself the hassle and just login as root, because you're circumventing basically every benefit sudo offers.
This is not really a problem with, or criticism of, the Registry, but one of the whole user-based security model (and one valid for rall major platforms, at least as they are typically used).
You could facilitate what you want by running each application in its own user account. A bit messy, but effective.
This way, removing registry entries would be easier.
Not easier, safer. The operation would be identical, there would be just less (if any) opportunity for bugs in the uninstallation process to modify other data.
Look, it's not worth arguing that the registry is a good design... we wouldn't be having this conversation if there was no problem.
I have never said the Registry was flawless. The basic design, however, is sound. It's certainly miles ahead of anything based around editing text files.
But I hope MS is breaking compatibility on this little gem in Vista, because if they don't, I'm not sure Vista will offer the ability to fend off badly written (and malicious)programs from hijacking the registry easily, just as they do in XP/2000/NT.
Programs could hijack textfiles jsut as easilly (more so, if anything).
The security in the Registry is excellent. Each individual key has per-user ACLs. It's much better security than plain text files are, or could practically be.
The *problem* - as you say later - is that lazy/incompetent/ignorant application developers write their applications badly.
This should be locked down just like the rest of Program Files, so admin/root is the only one who can make changes. Put per-user settings in Documents and Settings/username/Application Data.
This is exactly how the Registry works - system-wide setting go in HKEY_LOCAL_MACHINE and per-user settings go in HKEY_USERS\{SID}. Again, the problem is developers, not the Registry.
The only real problem with the Registry is that it is completely backed onto binary data files. Of course, when you consider it was specified and designed back in about 1990 when a 33Mhz 386 was a blazingly fast machine, that's an understandable design choice.
It's not just you - the whole thing pretty much screams out "troll".
The problem is not so much that NT's POSIX.1 layer was "fraudulent", it's that POSIX.1 on its own isn't particularly useful for anything fancy.
It's not just the "abuse" part that's vague, either. "Monopoly" is in no way formally and strictly defined, and a company really has no way of knowing that it is a monopoly until a court of law finds it to be one (at which point completely normal business practices suddenly become "monopoly abuse"),
Mainly because Linux and OS X - combined - still only represent around 1 in every 20-odd computers.
Even a "perfect storm" of Linux or OS X worms wouldn't have anything close to the impact of a relatively low-key Windows worm. There's simply not enough machines running either of them out there for it to spread especially fast or hit especially hard.
There have been *serious* flaws in OS X and Linux that have been patched since their release (more for Linux, obviously, since it's been around longer). Don't kid yourself otherwise. The reason these flaws have little noticable impact is because, relatively speaking, OS X and Linux have next to zero internet presence.
Try dropping a completely unpatched, default install of Redhat 7.x, 8.x or 9.x on a fast internet connection and see how long it lasts.
Legalese is required because the people writing it know they can't trust dishonest people (like them) to do the right thing - thus what they can and can't do needs to be spelt out in painfully explicit detail.
Suppose the poster is a business major who has been thrust into the IT/S division of his company, asking us business folk if he should have to learn these ridiculous technical terms in order to communicate with the people he has to deal with every day.
Difference is technical terms have real, concrete, well-defined meanings and are used to *clarify* discussion, whereas corporate-speak is typically the exact opposite.
I've yet to experience this sort of behaviour on any Windows system, although I've heard it bandied around frequently. I suspect it's much like that "Internet Explorer is part of the Windows kernel" that gets similarly parroted on Slashdot. Sure you're not repeating hearsay dating from NT 3.x and 4.0 ?
So "Windows, on well made hardware, with good drivers, politely used, is very stable"?
Yes. I can't even remember the time I saw a Windows NT machine crash where faulty hardware or bad drivers wasn't directly responsible (and easily identifiable as such).
Bollocks.
[...] or run any apps on it, [...]
More bollocks.
or use any devices (even well made ones) that use the same portion of the kernelspace at once.
Just like any other OS (assuming you're trying to say what I think you're trying to say).
The difference is that Windows doesn't do a good job of implementing levels of trust.
You don't appear to have the requisite technical knowledge to make that judgement.
What I don't expect is that they'll be able to bring down my OS, or any portion of it, when they do.
If you've never seen Linux or OS X kernel panic, you haven't been using either for very long.
Further, I expect that drivers written by Microsoft or that are Microsoft certified shouldn't have any compatibility problems. Otherwise, why did they get certified?
You may find this enlightening.
The USB subsystem shouldn't have the privileges to crash the rest of the OS. That's a design flaw.
Uh, it's a *hardware driver*. Hardware drivers (mostly) run in kernel space. Hardware drivers can most certainly crash the system.
This is not a design flaw, it's a design choice - one made for performance. Rest assured that faulty drivers are just as capable of bringing down Linux and OS X, as they are Windows.
This is just a single example that is typical of what you see in Windows, and the logic behind why I say that Windows itself is not stable. There are many, many others.
It's not typical behaviour at all. Your hardware is broken, and this is not Windows's fault.
Can you be a bit more specific ? What doesn't work ?
There are no shortage of stability problems with Linux when you get into the crappier end of the hardware scale, just like Windows (as I have been reminded over the last week losing more hair to machines with VIA chipsets).
MS isn't stable, because they don't care about stability
Windows on well made hardware, and with good drivers, is extremely stable. If your Windows installation is not stable, Windows is not the problem - it's either broken hardware or bad drivers.
The system design and development model has led to two things, a shortage of privilage escalation flaws (flaws isn't good enough, they have to allow a user account to gain root under conditions the virus can create) and a short lifespan of any such flaws that exist.
Very few pieces of malware need elevated privileges to do their "job".
grandchild.jpg.exe can never work on linux, period. You have to get the user to open a prompt cd to /home/granny/.email/files then chmod +X grandchild.jpg.exe, THEN ./grandchild.jpg.exe (in linux you have to create a launcher to execute a file in the gui, double-clicking will not work.
Users are happy to open password protected zipfiles to get at the empty promise of porn, viagra and free iPods, and you think a simple command like "chmod u+x file" is going to slow them down ?
s/system design/user behaviour/
No, processes started by another process inherit its privilege level
(This is so trivially simple to demonstrate that I'm amazed you'd even try and say it happens.)
"Luck" has nothing to do with it.
It's not FUD. It might not be the nicest way of putting it, but "viral" describes what the GPL does and was designed to do. Some GPLed code getting into a proprietry, commercial software codebase would have the potential to be very bad, especially if it happened deliberately.
You cannot "steal" something that is being given away.
Microsoft has done nothing but criticize and attack Open Source, but they'll exploit the efforts of that community.
Microsoft "criticise and attack" the *GPL*, not "Open Source".
However, they did it because it was convenient, not because it was morally right.
People develop under the BSD license so their code can be extensively distributed and reused. What is "immoral" about doing what they want ?