Frankly, in an enterprise setup I would be surprised if user home directories were not mounted with noexec (or whatever such an option would be called in Mac OS X), which would thwart this problem.
I wouldn't. The proprtion of Macs in the enterprise is miniscule to begin with, the proportion managed by highly-skilled admins (who would know about, and be able to implement, noexec) virtually nonexistent.
The vast, vast majority of Macs in the enterprise are "one off" purchases for C-level staff, or various other favoured individuals. They're rarely even noticed at by IT support, let alone actively managed.
If a cop pulls you over and you get out of the car without being asked, prepare for a world of hurt.
Which can lead to some pretty scary situations for tourists (or immigrants) from countries where you're _expected_ to get out of the car after being pulled over.
Everybody who can do math with variables can learn programming. It's one and the same, just with different notations. Programming is also a fundamental human skill. It's formalized planning, no more, no less. Finding the structure of a problem to be able to approach it systematically is something we do every day, even when we're not in front of a computer.
If the number of people who now have mortgages worth more than their homes, that they can't easily pay, won't convince you that not everyone can "plan" propertly, nothing will.
You point out a horribly flawed user interface that causes serious security problems as evidence that the underlying technology (already used on the iPhone and in SELinux among other places) can't work? If you present a user with a "Press OK to continue?" dialog on Windows, ever, you have failed in creating a user interface that will actually get the user to read and make a conscious choice. There are whole books on this interface failure.
What words you choose to put in the dialog box are a minor semantic issue. If the user is frequently bombarded with dialogs that require them to choose whether or not to continue, they *will* start simply hitting whichever button makes the dialog go away.
Ignorant user is right. Currently a user is not informed what access an app wants, in plain English, what the ramifications of that are, and then are not given any good choices about what to do. Gee, "something wants something can it do everything forever?" Brilliant! How about, "The application MacDefendor is from an unknown source and wants access to modify your Web browser and have complete control of your computer from now on. We recommend not allowing it this access. [Close MacDefendor] [Run MacDefendor, but restrict it to normal application privileges] [Allow MacDefendor complete control of my computer from now on].
As above, the exact message is semantics. The problem is the number and frequency of prompts. An additional problem with these sorts of prompts is that many situations simply can't be translated into "plain English" because they require non-trivial amounts of background understanding.
The problem with "Press OK to continue" dialogs - and make no mistake, that is what these are - is that they interrupt workflow, and lead to user fatigue such that they just start clicking whatever button is necessary to get the outcome they want.
You cannot secure any system, but you can do a better job than we do now for systems in general. Nor was I advocating, specifically, for an unmanaged system.
An "unmanaged system" is one where the ignorant end user has ultimate control. Ie: your typical desktop PC.
Personally I feel we should be breaking the application trust verification and sandbox preferences away from the application repository to introduce some competition.
That is to say, essentially the same situation computers have been in forever.
What is wrong with asking the user BEFORE allowing an app's sandbox to have these privileges?
1. You get the same "This program is going to delete all your data, send pictures of you with that asian hooker to your wife, list your house on eBay for $10, and kick your dog. Press OK to continue?" only multiplied by a hundred; and
2. If the ignorant end user has the ability to allow a program access, they will.
Yea, try managing 100s of windows desktops compared to 100s of linux desktops.
The "standard" (and I use the term loosely) centralised configuration tools for Linux are woefully bad.
Though most UNIX head don't realise this because they consider having to roll your own tools for solving common and ages old problems to be a good thing, rather than a complete waste of their time.
If there is anything for Aussies to be ashamed of here, it's our media. Fairfax is using the non-event to talk itself up and create a mountain out of a molehill. I've always said, Australian media was better then US media but today, they've made a liar out of me.
It arguably still is, but the same people who made the US media as bad as it is today, are hard at work in Australia trying to fix the discrepancy.
If you went a flat tax, then yeah, low income people would suffer at first, but I'll tell you what, there would be no greater motivation for them to get off their ass and improve their life/situation/job/whatever.
Er, as opposed to now ?
If you earn enough to be sitting in the top income tax bracket, the extra tax you're paying is not going to be a meaningful disincentive to further effort. Especially in a relatively low-taxing country like the US.
Weird. I always thought it was the kernel and kernel helpers it ran in ring 0 - with the drivers (EXCEPT HPFS386) run in Ring 2 or ring 3 (depending on the driver layer or driver).
Rings 2 and 3 are used for user mode code - privileged and unprivileged, respectively.
Anyway, IBM also managed a kernel for entirely different architecture - namely the PPC and it's different architecture. How odd that Microsoft couldn't manage such.
Windows NT is, or has been, ported to MIPS, PPC, Alpha, SPARC, PA-RISC, x86-64 and ARM.
The simple fact is that Microsoft made the same architectural choices that basically everyone else in the same position did, for the same reasons.
Ah... I see... IBM, in little time, managed to rectify such a situation, but in nearly 2 decades, Microsoft has not?
IBM weren't writing a portable OS.
Moving parts of (or entire) drivers into ring 0 that don't belong there (Rings 1-2 were intended for that, as you probably already know), moving other Windows code there (c'mon, just fix the damn speed problems... don't try to make Windows faster by trying to minimize ring/mode switching - OS/2 handled CONTINUAL ring switching with far better speed than the NT line, even with the NT line minimizing such)...
What "doesn't belong there" ? Every other x86 OS - including OS/2 - ran drivers in ring 0 to maintain acceptable performance levels.
...even though other architectures did support similar, as is evidenced by operating systems on such architectures taking advantage of such capabilities. Odd...
Not true... OS/2 1.x didnt support them, did they? When OS/2 2.x onwards was written/rewritten by IBM for 386 and better hardware, they seem to have managed fine - nor did they just simply "tack it on".
OS/2 1.x and 2.x are completely different codebases. Windows NT and, well, Windows NT are not.
And regardless, that does not explain Microsoft moving things into the incorrect ring(s). If they can move stuff into the wrong rings, why couldn't they move them into the correct ring(s) instead?
He's correct. The only "mainstream" (past/present) OS that actually utilized the CPU's protection levels to any decent extent was OS/2 - which is also why it was a bitch to run in numerous virtual machines (most notably due to poor virtual Ring 2 support). And in reality, Microsoft glomming a whole bunch of things into Ring 0 is a step backwards.
Windows NT only uses two rings because that's all some of its historically support platforms had. I imagine it's something that's rather difficult to tack on later.
To a point. When you use your massive war chest to create a product that you give away for free just to cut revenue of a competitor (to put them out of business) then antitrust violation of the law is up to judges to interpret.
Note that there is not a single feature in Windows - or any other modern OS - this does not describe.
Used to be you had to pay for everything from a text editor to a TCP/IP stack and you had to glue the whole shebang together yourself like a patchwork quilt.
if it hadn't been for this anti-trust case, Microsoft would have crushed Apple like a bug, just like it did all it's other competitors before it.
Nope. Apple's revival was already beginning (ie: the first iMac) as the antitrust suit was filed.
Anyone remember Wordperfect? Do you remember the guys who invented the spreadsheet? Anyone remember the company who invented visual programming? Anyone remember the company that put out the first commercial web browser? Anyone remember GEOS? BeOS??
Yes. All beaten out by products that customers preferred.
Instead, Microsoft had to actively support Apple, including the massive investment in porting Office to Mac, release after release, even through Apple's transition to a BSD-like subsystem. Why? Because Microsoft didn't want to get sued again.
Apple and Microsoft don't even compete in the same market. Apple's survival (or lack thereof) would have had zero impact on Microsoft's monopoly status.
But, IF - and I stress IF - Binny was shot down like a rabid dog, then the US should have announced it in just those terms. There's no need to pull punches, gloss over the truth, or to sugar coat it. Just tell the world, "We killed the bastard, end of story!"
A snatch & grab "gone wrong" - especially on someone else's sovereign territory - is easier to justify diplomatically and legally, than a hit squad.
The problem with regulation by lawsuit is that the legal system is slow and clumsy.
The vastly more important problem is that instead of preventing problems from occurring in the first place, it merely offers a path of restitution for those whom bad things have happened to.
I'm not sure why you are so against libertarianism when the current model is leading to these terrible outcomes, and despite huge amounts of effort, time and money invested, the situation never seems to improve.
Because before we had the current model, the outcomes were even worse.
I am simultaneously amused and despaired that you think "the little guy" would have any chance of victory in a lawsuit against a giant mining corporation, perplexed how you think your model could possibly manage risk pro-actively, and outraged you think it is a reasonable response to, say, someone being killed.
Slashdot Mirror
I wouldn't. The proprtion of Macs in the enterprise is miniscule to begin with, the proportion managed by highly-skilled admins (who would know about, and be able to implement, noexec) virtually nonexistent.
The vast, vast majority of Macs in the enterprise are "one off" purchases for C-level staff, or various other favoured individuals. They're rarely even noticed at by IT support, let alone actively managed.
Which can lead to some pretty scary situations for tourists (or immigrants) from countries where you're _expected_ to get out of the car after being pulled over.
If the number of people who now have mortgages worth more than their homes, that they can't easily pay, won't convince you that not everyone can "plan" propertly, nothing will.
Safari is "integrated" into OS X the same way IE is into Windows.
As opposed to paying it to a landlord (who, given Australia's current state of property "investment" is just giving it to the bank anyway) ?
What words you choose to put in the dialog box are a minor semantic issue. If the user is frequently bombarded with dialogs that require them to choose whether or not to continue, they *will* start simply hitting whichever button makes the dialog go away.
As above, the exact message is semantics. The problem is the number and frequency of prompts. An additional problem with these sorts of prompts is that many situations simply can't be translated into "plain English" because they require non-trivial amounts of background understanding.
The problem with "Press OK to continue" dialogs - and make no mistake, that is what these are - is that they interrupt workflow, and lead to user fatigue such that they just start clicking whatever button is necessary to get the outcome they want.
An "unmanaged system" is one where the ignorant end user has ultimate control. Ie: your typical desktop PC.
That is to say, essentially the same situation computers have been in forever.
1. You get the same "This program is going to delete all your data, send pictures of you with that asian hooker to your wife, list your house on eBay for $10, and kick your dog. Press OK to continue?" only multiplied by a hundred; and
2. If the ignorant end user has the ability to allow a program access, they will.
You cannot secure an unmanaged system.
The "standard" (and I use the term loosely) centralised configuration tools for Linux are woefully bad.
Though most UNIX head don't realise this because they consider having to roll your own tools for solving common and ages old problems to be a good thing, rather than a complete waste of their time.
It arguably still is, but the same people who made the US media as bad as it is today, are hard at work in Australia trying to fix the discrepancy.
Er, as opposed to now ?
If you earn enough to be sitting in the top income tax bracket, the extra tax you're paying is not going to be a meaningful disincentive to further effort. Especially in a relatively low-taxing country like the US.
Uh, isn't the ultimate stated goal of a rather large proportion of Islamic terrorists the establishment of a global Caliphate ruled by Sharia law ?
Yes
Rings 2 and 3 are used for user mode code - privileged and unprivileged, respectively.
Windows NT is, or has been, ported to MIPS, PPC, Alpha, SPARC, PA-RISC, x86-64 and ARM.
The simple fact is that Microsoft made the same architectural choices that basically everyone else in the same position did, for the same reasons.
If the end user can do it, the end user can be convinced to do it by malware.
IBM weren't writing a portable OS.
What "doesn't belong there" ? Every other x86 OS - including OS/2 - ran drivers in ring 0 to maintain acceptable performance levels.
For example ?
OS/2 1.x and 2.x are completely different codebases. Windows NT and, well, Windows NT are not.
By which you mean...?
You mean they made the same engineering decision basically every other OS vendor did ?
(Not to mention at least on of Windows NT's historically supported platforms only had two rings, which is why that's all NT uses.)
Windows NT only uses two rings because that's all some of its historically support platforms had. I imagine it's something that's rather difficult to tack on later.
Indeed. If it's on the Internet, it must be true !
Huh ?
The commonly accepted "wisdom" on Slashdot is that marketshare is irrelevant. Ergo, infection rates should not change.
Note that there is not a single feature in Windows - or any other modern OS - this does not describe.
Used to be you had to pay for everything from a text editor to a TCP/IP stack and you had to glue the whole shebang together yourself like a patchwork quilt.
Do you _really_ think things were better then ?
Nope. Apple's revival was already beginning (ie: the first iMac) as the antitrust suit was filed.
Yes. All beaten out by products that customers preferred.
Apple and Microsoft don't even compete in the same market. Apple's survival (or lack thereof) would have had zero impact on Microsoft's monopoly status.
A snatch & grab "gone wrong" - especially on someone else's sovereign territory - is easier to justify diplomatically and legally, than a hit squad.
The vastly more important problem is that instead of preventing problems from occurring in the first place, it merely offers a path of restitution for those whom bad things have happened to.
Which isn't much if the "bad thing" killed you.
Because before we had the current model, the outcomes were even worse.
I am simultaneously amused and despaired that you think "the little guy" would have any chance of victory in a lawsuit against a giant mining corporation, perplexed how you think your model could possibly manage risk pro-actively, and outraged you think it is a reasonable response to, say, someone being killed.