What would be the point of such a machine? The GMA 950 is a perfectly adequate chip for everything but games. And for games, well, if you're playing games, why the hell are you thinking of buying a Mac? Doom III is still considered a "new release" in the Mac world!
I haven't seen any evidence of lower prices on the Dell. Apple has kept the same price points, but they've gradually upgraded to faster Core 2 Duos over that period. A MacBook with 2 GHz Core 2 and 1 GB of RAM costs $1300. A Latitude D620 with the same basic features (1 GB RAM, 80 GB HD, DVD burner, wifi and bluetooth) costs $1268, with Dell's "$366 limited time instant savings". The Dell is all of $32 cheaper, and doesn't have Firewire, a webcam, etc.
Nice rewriting of history. All the indications are that PA Semi was very surprised when Apple didn't pick them as a supplier for notebook hardware. They were expecting Apple to be a major competitor. Also, if IBM had planned on dumping Apple, do you really think they would've added VMX (AKA Altivec) to the Power6 architecture? Since it only supports single-precision FP, and the chip has dual double-precision FMAC units anyway, the VMX unit is almost completely useless except for compatibility with OS X software that uses AltiVec.
And finally, the Intel Macs are broadly price competitive. Even months after its release, the only notebook that competes with the MacBook in its segment is the Vaio C series, and the latter is the same price for fewer features. And of course the Mac Pro is an absolute steal for a quad Xeon machine, even all this time after its release with no updates.
You fail to understand the issue here. In this context, "stuff not working" doesn't necessarily mean "Y does it already but we need to fix X to do it", but rather "neither X nor Y do it, and we have to implement it in one of them". Consider companies like Broadcom or Atmel or Sony. They do Linux work either to create drivers for their new hardware, or to port Linux to their new platform. Sony, for example, did a lot of work on the last couple of releases because they added support for their Cell platform. Interestingly, that's why a lot of these companies do their work on Linux in the first place. It costs Sony or Broadcom a lot less to port Linux to their new platform than it would to work out a licensing agreement for Windows and then port that!
Actually, this shows how *well* open source software works. Here you have a whole bunch of big companies investing a lot of money into producing honest-to-goodness copyleft free software. RMS was absolutely right --- software can be successful without being burdened by a license that takes rights away from its users.
Uh, yes? Christmas day is actually a very busy time for movie theaters, since everybody is off work and lots of people go to see movies with their family after Christmas eve celebrations the night before. It's quite common for a studio to release a big budget movie on Christmas day. All three LOTR movies were released on Christmas day, for example.
And how many potential geniuses do we miss out on when we teach 50% of our population to prioritize making babies over perusing their talents and goals?
If features are so compelling, why does the iPod have more market share than any of its competitors, almost all of which have more features (integrated FM, etc)?
As for the Newton, it failed because it was a poorly executed product, and the Pilot was a very well-executed one. The reason Palm is even alive to this day is because of the people who prefer its interface to Windows Mobile.
It's just the nature of human history. Empires rise and fall, cultures go through peaks and valleys. Civilizations in India had public sewer systems in major urban centers 4000 years ago, at a time when the people of England didn't even have cities. South American Indians built great urban civilizations at a time when Europeans were still in the Dark Ages. For a period of about 500 years during the Muslim Empires, there was a flourishing of Islamic civilization. In the context of its time, the Islamic Empires were quite advanced, and liberal as far as empires go. The treatment of Jews and Christians by the Muslims that conquered Spain (taxes, discrimination) would not fly in a modern society, but were a sure sight nicer than how the Spanish Christians treated the Jews and Muslims 700 years later during the Inquisition!
Now, as for what happened in this specific case. At the point where Europe started blossoming, and the colonial powers started their rise, the civilized Islam world was already half a millennium old. It was in a period of decay, caused by the usual suspects (political infighting and fragmentation, lack of strong leadership, overextension of power and resources, etc). By the time the colonial powers came into their own, they were able to mop up the remaining Islamic powers (eg: Ottoman Empire) that hadn't already self-destructed. With the demise of civilization came poverty, and with that came all the ills of poverty (conservatism, paranoia, a general reversion to baser practices). What you see in much of the Islamic world today is the kind of "Lord of the Flies" reversion of formerly civilized people to their more basic nature when they are removed from civilized society.
It is important to note that there are remnants here and there of civilized Islamic society. For example, the urban centers of Iran have always been highly civilized places, going back thousands of years. Iran might be highly antagonistic to the West, and there is a contingent of the society that uses religion as a tool for controlling the population, but the basic elements of civilization are undeniably strong. The balance between theocracy and democracy in their government is complex and constantly shifting, but the people exercise very real power in their governance. Assuming the current unfortunate stand-off with the West doesn't end badly, there is a very good chance that Iran will become a fully modern state in the next fifty years.
It's not the disk space, it's the memory usage. Loading only a single instance of a library would work, but if it was an automated system, you'd have to be conservative and basically only perform that optimization when the libraries were bit-for-bit identical. Without the organizational structure provided by packagers, it'd be very unlikely that this would ever be the case, since even changing the compiler flags slightly would create a separate library. And this would apply to almost every library, because in the Linux world, most libraries are third party libraries. It's not like OS X or Windows, where one company develops the whole stack.
1) "Usual", not "useful". It's like how it's not "usual" for OS X users to use Java apps, so it's not really fair to throw that in there.
2) You're missing the point. FireFox and OpenOffice use a custom toolkit on every platform. So they cancel out in the equation here.
3) No they don't. Core Foundation is a very low level infrastructure library, not really a part of the toolkit. Both are being CF-ized, but that doesn't amount to much. Cocoa and Carbon are joined much lower than the widget level. As of 10.4, they're joined (mostly) at the HIView layer, which is conceptually a little bit higher up than X11 (an HIView "view" roughly corresponds to a nested X11 window, with a bit more functionality). Above that, they're still independent toolkits, with independent widgets (aside from window-level widgets like menus, dialogs, etc), independent behaviors, etc.
You have to keep in mind that Cocoa and Carbon were originally completely separate toolkits. The former is OpenStep, while the latter is derived from the Quicktime for Windows portability layer. The former was built on DPS, and migrated to Quartz, while the latter was built on QuickDraw. Apple has been gradually been merging them from the bottom-up. Eventually, both will just be API wrappers over HIView/HIToolbox/Quartz, but that's some time away yet.
There is also another strength of package managers that the app-bundle system lacks. In OS X, apps will pester you to update them. Colloquy, which releases several times a month, gets really annoying in this regard. In OS X, apps have to do this because there is no way to update all apps automatically. A package-managed system doesn't have this problem.
There are substantial deficiencies with OS X-style packages, which would only be amplified in the Linux world. In general, OS X packages have to bundle libraries that are not included by default in the OS. This leads to anywhere from a little to a lot of wasted disk space and memory. The problem is limited in OS X, because there is a standard set of frameworks Apple ships with, but this is not true on Linux. As a result, app bundles would have to include things like GNOME as well.
1) It's not terribly usual for GNOME users to use KDE apps, so you're down to 3 widget sets there. 2) Firefox and OpenOffice define their own widget sets on every system they run on. If you use Camino and NeoOffice, there's still XUL and VCL under the hood. 3) There is no way to get an OS X desktop with only one widget set either. Carbon and Cocoa are always there.
In theory, caching the permission does open up a hole. In practice, the risk is probably not that substantial. If the file permissions mechanism is working, the executable can't be compromised by modifying its binary. So the only attack vector is modifying its running image. There are various layers of protection (not the least of which is the fact that the application is running in a protected address space) that makes this attack complicated. And of course in a well designed system, only a few applications (file manager, configuration application) would ever need to run with elevated permissions, and you'd hope they'd be written to be resistant to compromise.
The risk with caching the permission is probably comparable to 'sudo' or the various 'setuid root' binaries that run on a UNIX system (like the X server). They're definitely something to watch out for in a highly-secured installation, but in practice they don't seem to cause many problems in desktop environments.
It's not a matter of the "app in question" being compromised. Vista doesn't elevate the entire user, it only elevates the application.
Right, which is why I referred to the "app in question".
For the entire length of execution of that application, the application will run elevated. For instance, Visual Studio.NET is an application that pretty much always needs to be run as admin.
Speaking of spectacularly bad ideas...
If you cached the elevated credentials authorization for "X" minutes, or whatever, you would be giving a free pass to any malware that happened to be trying to do something bad.
The elevated permissions would apply per-process. The app would request permission, do something, and drop the permissions. If it requested it again within the timeout, it'd get the permission again without a prompt.
No, it doesn't. By default, all users on the system can read files in c:\Program Files.
The hell it doesn't.
It does throw up a UAC, but I've already explained why the "caching period" is a bad idea.
You misunderstood the idea of the timeout. It would indeed be inane to elevate the user for the entire duration. Note that 'sudo' doesn't do that.
The better solution is what OS X does: extend "sudo" to the GUI. The first time the app needs escalated privileges, prompt for the user's password. Then, cache those privileges for a reasonable amount of time and don't prompt. Unless the app in question is compromised in that interval, it doesn't matter.
The problem with UAC is that it fails to separate the two orthogonal issues of sanity-checking the user's behavior, and maintaining system security. Consider how "Program Files" is handled. Browsing into "Program Files" throws up a UAC alert. It shouldn't do that --- "Program Files" is readable to everyone. Writing to "Program Files" should throw up a UAC alert, but only the first time in the caching period. The question at that point isn't "Do you really want to modify this directory" (of course I do!), but rather "Do you want to give Explorer.exe permission to modify this directory". When you follow the first train of thought, you end up with prompting the user each time, because obviously each copy requires a separate sanity-check. If you follow the second train of thought, you see that the caching mechanism is just fine, since if Explorer.exe was authorized 30 seconds ago, it's unlikely it was compromised since then, and should retain that authorization.
The "next, next, next" type install has a lot of pitfalls. First of all is the whole idea of an installer program, which is certainly not an intuitive concept. It's one thing to start a program to browse for other programs (Synaptic), or to have a program just be a bundle you can download from a website, but to have a program that you download actually be a proxy to install the program you really want? Not obvious. Then there is the fact that most inexperienced people see all the options in the installer and freak. For most software, none of those options matter! That's why you people derisively call them "Next, next, next" installers, because you can just mash the "Next" button the whole time and get the correct result. Aside from a few rare, big applications (where choosing components is important), a simple "install the fucking app already" button is all you really need.
Most of those prompts were redundant, either because they enforce things guaranteed by the underlying file permissions, or because the authorization could've been cached.
Vista:Are you sure you want to run this program?
Of course! It's got +X set!
Vista: are you sure you want to run this? it's a program, you know.
Ditto.
Vista: This is from AMD. Do you trust AMD?
Redundant. If I didn't trust them, I wouldn't have set +X.
Vista: Are you sure you want to go there?
Since Program Files shouldn't be world writable, this should prompt you for the administrator password. This authoriation should then be cached for Explorer.exe.
Vista: Are you sure you want to overwrite this file?
I'll let this slide, because even 'cp' prompts for that.
Vista:A program wants to write to the Program Files folder. Is this ok?
Should've grabbed cached authorization for Explorer.exe. Unless Explorer.exe was compromised in the 30 seconds between this action and the previous one, no security is lost here.
Vista:You are trying to copy from a network share to the program files folder. This isn't allowed. Hit ok.
That's just idiotic.
Are you sure you want to overwrite this file?
Again, I'd let it slide depending on preference.
Vista:A program wants to write to the Program Files folder. Is this ok?
Cached authorization again.
It's really not that hard. UNIX/sudo got this right god knows how long ago. Apple did the right thing and just copied the sudo mechanism wholesale. Microsoft should to.
How many grandmas can figure out a Windows "next, next, next" type install?
Especially since you can use the "Add/Remove" programs applet in Ubuntu to (gasp!) actually add programs (unlike in Windows), your rant is kind of hollow.
Or maybe they're taking the money saved on the "slower" RAM (that isn't any slower in actual use) and spending it on better motherboards or PSUs ?
Which one do you really thing is more likely? Honestly? If they're trying to save a few bucks on something like RAM, which is a spec they can advertise, what do you think they're doing with the stuff they can't advertise. I've been around too many Dells in the last few years to give them the benefit of the doubt.
Seriously, it's a non-issue. Dell still have options for 533Mhz DDR2 on machines like their PowerEdge 2950s, which aren't quite solid and reliable servers.
The 2950 uses server-grade, high-capacity FB-DIMMs, which are a whole different kettle of fish from what they're putting in the E520! When you see a lower speed rating on a server DIMM that costs $3000, you assume conservativeness, not cost-cutting. I have a hard time believing the 533 MHz DIMMs in the E520 are otherwise high-quality parts that are just conservatively rated. More likely they are low-quality parts sold at a discount to get rid of the inventory of 533 MHz chips.
Apple do the case design and maybe a bit of internal layout to help get things into the case they want, but now they've switched to Intel pretty much everything is off-the-shelf and built in the same factories pumping out regular old PC components - because they *are* regular old PC components.
Sure, Mac's use a lot of the same parts as PCs. However, Apple still has control over parts selection, and still custom-designs their motherboards, power supplies, cases, and cooling systems. In my experience with building PCs, these are the reliability bottlenecks. I rarely see PCs where the Intel CPU or chipset craps out. On the other hand, I quite often see PCs where the motherboard is a cheap 3-layer piece of shit, or the "500W" power supply can't reliably deliver half that, or where the case is made of cheap plastic and has flimsy bits that break off easily. You're not going to find that stuff in an Apple machine. Of course, you can find it in the PC world too (think ThinkPads, or building it yourself), but then you're going to have to be prepared to spend as much money as you would on an Apple machine.
Added to that, even if they did do the whole design top to bottom, that still wouldn't impact build quality. You can have the best architect in the world, but if the builders and incompetent and the cement bad, the building is still going to fall down.
The build quality is largely a factor of how much you're willing to spend. You can easily have 0.01" tolerance on a spec, instead of 0.1" tolerance, even with the same builder, it's just going to cost you more money. Dell's suppliers in Taiwan are probably easily capable of building a machine with the same tolerances as an Apple, but it seems that since about 2000 or so, Dell isn't willing to pay for that.
iMacs (and Mac Minis and MacBooks) are built to a budget. Mac Pros (and the other "pro" machines aren't (well, are less so), and carry a suitable price premium to compensate (and they _are_ very nice inside - well, the MBPs are a bit of a mess...). You can get the same thing from Dell, as well, if you want to spring for a Precision workstation or higher-end Latitude (vs Inspiron) notebook (or go IBM/Lenovo).
Every machine is build towards a target price. However, Apple seems to be less willing to trade user experience for that target price than does Dell. Even the MacBook has the same close-flush design that causes Apple's cases to cost more to build. Even the iMac uses high-quality acrylics and hard plastics that cost more than the softer stuff you see in most Dells. Apple could've saved quite a bit of money on the iMac by making it an inch thicker, specifying a more powerful (and hence louder) fan, and using desktop parts instead of laptop ones. They didn't because they didn't want to trade off the user experience.
To be fair, Apple isn't the only one who builds machines like this. Lenovo's Th
see, this is the problem with a mac fan, you think that someone actually wants the parts just how apple give them. I know its hard for you, but try to imagine this isn't my first computer and most of those gadgets I already own.
If you don't want the parts, then fine, don't buy the machine. But don't say you can get a similar machine for a lot less money (which is what you did!) If you want to compare price, then spec out a roughly comparable machine, and compare the price. If you want to argue about flexibility, then argue that instead. I'm perfectly happy to admit that Apple's canned configurations are nowhere near as flexible as building something yourself, that's really a no-brainer.
now, even after buying a second monitor, I have over 600 bucks left over. I'm comparing with the tower now, which is at least 2500 bucks w/o monitor, much less dual monitors. I'm quote happy with my monitor so I'd get the same one for a mac tower if I had bought one. Just the box, I'm at half the price, far more powerful than any iMac I could have gotten, and with my use of the computer, close enough to a mac tower at the time.
The Mac Pros are just about six months old. They're all quad core (2 x Xeon), and always have been. They're completely out of the league of your machine. In addition to the pair of Xeons, they've got a server-grade Intel 5000X motherboard that'll run you $400-$500 by itself. They've got quad-channel FB-DIMM memory and 8 memory expansion slots. And the parts alone will cost you about as much as a Mac Pro, even neglecting labor, warranty, etc.
Firstly, 533Mhz DDR2 isn't "last-gen tech". DDR1 is "last gen tech". There are numerous machines out there still only using 533Mhz RAM - and most of the ones using 667Mhz RAM derive no benefit except in a handful of synthetic benchmarks.
The issue with DDR2-533 is that it's kind of a bastard-stepchild speed grade. It had a run of only about a year, after being introduced with the earliest Intel DDR2 platforms and supplanted by DDR2-667 very shortly afterward. DDR2-667 has been common now for almost two years, and as you said the price difference is minimal. Since all Core 2 chipsets support DDR2-667, the only reason to ship the older memory in a machine would be to save a few extra cents. It's that sort of scrimping that worries me. After all, if they don't want to pony up a couple of extra bucks for RAM, who knows how much they're skimping on the PSU or motherboard.
The iMac is assembled to a budget - to save "every last penny" - and with the cheapest parts available to fill its requirements (which happen to include a particularly small form factor, requiring some low-power parts). Don't kid yourself otherwise. Like I said, iMacs roll off the same assembly lines PCs like the E520 do. Heck, MacBook Pros are just rebadged Asus laptops (or certainly used to be).
This is actually very much incorrect. Yes, the same Taiwanese manufacturers (like Asus) actually build the machines, but Apple is the one that does the design (no Apple laptop is a re-badged anything). Moreover, it's well known in the industry that Apple's tolerances are a lot tighter than everyone else's. Take their laptops for example. The MacBooks has a case design that is really hard to build. When closed, you can run your finger along the seam between the lid and the body, and it should be tight and flush. Even a slight misalignment results in a noticeable defect. As a result, they have to specify stricter tolerances on the build, which costs money. Almost every other laptop design includes some sort of rounded lip, which can hide substantial misalignment without the user noticing, and can thus be manufactured more cheaply. Also, take a look inside a PowerMac or Mac Pro sometime. Apple doesn't skimp on things like the PSU, motherboard, or even case fans. A lot of manufacturers do, because a great PSU is not something they can advertise in the spec sheet. Apple's PowerMac PSUs run from 700W to 1000W (real, not like the $30 "500W" PSUs you see on NewEgg), and they're heavy and solid like a PSU should be. The PowerMac (and probably Mac Pro)'s fans are a custom variant of a Delta model that runs $10-$12 on the internet. You're not going to find a $10+ case fan in that E520.
Note that Dell used to build machines like this too. I've got a Dimension D300 from 1998, and its probably as heavy as a modern E-series machine despite being substantially smaller. It uses more expensive hard plastics (as opposed to the cheaper soft plastics you see on modern Dells), and an Intel Seattle motherboard modified for better power delivery. The damn thing is getting close to 10 years old, and is still running flawlessly. A couple of years ago it was sitting in a closest acting as a server while the room was remodeled. I realized after a couple of months that it had been inhaling all the dust from the remodeling for all that time, yet it never even so much as crashed.
Now, Apple certainly isn't as premium as you can get. I've spent $20+ on a CPU fan because that has a drastic effect on the noise profile of the machine. I've spent $130 on an Antec P180 case because of the noise-absorbing aluminum/plastic/aluminum sandwich paneling. These are a cut above even what you find in an Apple machine. However, I've dug around my fair share of cheaper PCs, and almost all of them cut corners on these sorts of parts.
You seem to have that arse-about-face. Gaming - on of the few things the average person will do that actually stretches a modern computer - is something where you most certainly can *not* get away with "a really shitty machin
Slashdot Mirror
What would be the point of such a machine? The GMA 950 is a perfectly adequate chip for everything but games. And for games, well, if you're playing games, why the hell are you thinking of buying a Mac? Doom III is still considered a "new release" in the Mac world!
I haven't seen any evidence of lower prices on the Dell. Apple has kept the same price points, but they've gradually upgraded to faster Core 2 Duos over that period. A MacBook with 2 GHz Core 2 and 1 GB of RAM costs $1300. A Latitude D620 with the same basic features (1 GB RAM, 80 GB HD, DVD burner, wifi and bluetooth) costs $1268, with Dell's "$366 limited time instant savings". The Dell is all of $32 cheaper, and doesn't have Firewire, a webcam, etc.
Nice rewriting of history. All the indications are that PA Semi was very surprised when Apple didn't pick them as a supplier for notebook hardware. They were expecting Apple to be a major competitor. Also, if IBM had planned on dumping Apple, do you really think they would've added VMX (AKA Altivec) to the Power6 architecture? Since it only supports single-precision FP, and the chip has dual double-precision FMAC units anyway, the VMX unit is almost completely useless except for compatibility with OS X software that uses AltiVec.
And finally, the Intel Macs are broadly price competitive. Even months after its release, the only notebook that competes with the MacBook in its segment is the Vaio C series, and the latter is the same price for fewer features. And of course the Mac Pro is an absolute steal for a quad Xeon machine, even all this time after its release with no updates.
You fail to understand the issue here. In this context, "stuff not working" doesn't necessarily mean "Y does it already but we need to fix X to do it", but rather "neither X nor Y do it, and we have to implement it in one of them". Consider companies like Broadcom or Atmel or Sony. They do Linux work either to create drivers for their new hardware, or to port Linux to their new platform. Sony, for example, did a lot of work on the last couple of releases because they added support for their Cell platform. Interestingly, that's why a lot of these companies do their work on Linux in the first place. It costs Sony or Broadcom a lot less to port Linux to their new platform than it would to work out a licensing agreement for Windows and then port that!
Actually, this shows how *well* open source software works. Here you have a whole bunch of big companies investing a lot of money into producing honest-to-goodness copyleft free software. RMS was absolutely right --- software can be successful without being burdened by a license that takes rights away from its users.
Apparently, I can't read or write. LOTR did not open Christmas day. They're among the movies with the highest Christmas-day gross ticket sales.
Uh, yes? Christmas day is actually a very busy time for movie theaters, since everybody is off work and lots of people go to see movies with their family after Christmas eve celebrations the night before. It's quite common for a studio to release a big budget movie on Christmas day. All three LOTR movies were released on Christmas day, for example.
And how many potential geniuses do we miss out on when we teach 50% of our population to prioritize making babies over perusing their talents and goals?
If features are so compelling, why does the iPod have more market share than any of its competitors, almost all of which have more features (integrated FM, etc)?
As for the Newton, it failed because it was a poorly executed product, and the Pilot was a very well-executed one. The reason Palm is even alive to this day is because of the people who prefer its interface to Windows Mobile.
It's just the nature of human history. Empires rise and fall, cultures go through peaks and valleys. Civilizations in India had public sewer systems in major urban centers 4000 years ago, at a time when the people of England didn't even have cities. South American Indians built great urban civilizations at a time when Europeans were still in the Dark Ages. For a period of about 500 years during the Muslim Empires, there was a flourishing of Islamic civilization. In the context of its time, the Islamic Empires were quite advanced, and liberal as far as empires go. The treatment of Jews and Christians by the Muslims that conquered Spain (taxes, discrimination) would not fly in a modern society, but were a sure sight nicer than how the Spanish Christians treated the Jews and Muslims 700 years later during the Inquisition!
Now, as for what happened in this specific case. At the point where Europe started blossoming, and the colonial powers started their rise, the civilized Islam world was already half a millennium old. It was in a period of decay, caused by the usual suspects (political infighting and fragmentation, lack of strong leadership, overextension of power and resources, etc). By the time the colonial powers came into their own, they were able to mop up the remaining Islamic powers (eg: Ottoman Empire) that hadn't already self-destructed. With the demise of civilization came poverty, and with that came all the ills of poverty (conservatism, paranoia, a general reversion to baser practices). What you see in much of the Islamic world today is the kind of "Lord of the Flies" reversion of formerly civilized people to their more basic nature when they are removed from civilized society.
It is important to note that there are remnants here and there of civilized Islamic society. For example, the urban centers of Iran have always been highly civilized places, going back thousands of years. Iran might be highly antagonistic to the West, and there is a contingent of the society that uses religion as a tool for controlling the population, but the basic elements of civilization are undeniably strong. The balance between theocracy and democracy in their government is complex and constantly shifting, but the people exercise very real power in their governance. Assuming the current unfortunate stand-off with the West doesn't end badly, there is a very good chance that Iran will become a fully modern state in the next fifty years.
It's not the disk space, it's the memory usage. Loading only a single instance of a library would work, but if it was an automated system, you'd have to be conservative and basically only perform that optimization when the libraries were bit-for-bit identical. Without the organizational structure provided by packagers, it'd be very unlikely that this would ever be the case, since even changing the compiler flags slightly would create a separate library. And this would apply to almost every library, because in the Linux world, most libraries are third party libraries. It's not like OS X or Windows, where one company develops the whole stack.
1) "Usual", not "useful". It's like how it's not "usual" for OS X users to use Java apps, so it's not really fair to throw that in there.
2) You're missing the point. FireFox and OpenOffice use a custom toolkit on every platform. So they cancel out in the equation here.
3) No they don't. Core Foundation is a very low level infrastructure library, not really a part of the toolkit. Both are being CF-ized, but that doesn't amount to much. Cocoa and Carbon are joined much lower than the widget level. As of 10.4, they're joined (mostly) at the HIView layer, which is conceptually a little bit higher up than X11 (an HIView "view" roughly corresponds to a nested X11 window, with a bit more functionality). Above that, they're still independent toolkits, with independent widgets (aside from window-level widgets like menus, dialogs, etc), independent behaviors, etc.
You have to keep in mind that Cocoa and Carbon were originally completely separate toolkits. The former is OpenStep, while the latter is derived from the Quicktime for Windows portability layer. The former was built on DPS, and migrated to Quartz, while the latter was built on QuickDraw. Apple has been gradually been merging them from the bottom-up. Eventually, both will just be API wrappers over HIView/HIToolbox/Quartz, but that's some time away yet.
There is also another strength of package managers that the app-bundle system lacks. In OS X, apps will pester you to update them. Colloquy, which releases several times a month, gets really annoying in this regard. In OS X, apps have to do this because there is no way to update all apps automatically. A package-managed system doesn't have this problem.
There are substantial deficiencies with OS X-style packages, which would only be amplified in the Linux world. In general, OS X packages have to bundle libraries that are not included by default in the OS. This leads to anywhere from a little to a lot of wasted disk space and memory. The problem is limited in OS X, because there is a standard set of frameworks Apple ships with, but this is not true on Linux. As a result, app bundles would have to include things like GNOME as well.
Your example is wrong on two fronts:
1) It's not terribly usual for GNOME users to use KDE apps, so you're down to 3 widget sets there.
2) Firefox and OpenOffice define their own widget sets on every system they run on. If you use Camino and NeoOffice, there's still XUL and VCL under the hood.
3) There is no way to get an OS X desktop with only one widget set either. Carbon and Cocoa are always there.
In theory, caching the permission does open up a hole. In practice, the risk is probably not that substantial. If the file permissions mechanism is working, the executable can't be compromised by modifying its binary. So the only attack vector is modifying its running image. There are various layers of protection (not the least of which is the fact that the application is running in a protected address space) that makes this attack complicated. And of course in a well designed system, only a few applications (file manager, configuration application) would ever need to run with elevated permissions, and you'd hope they'd be written to be resistant to compromise.
The risk with caching the permission is probably comparable to 'sudo' or the various 'setuid root' binaries that run on a UNIX system (like the X server). They're definitely something to watch out for in a highly-secured installation, but in practice they don't seem to cause many problems in desktop environments.
It's not a matter of the "app in question" being compromised. Vista doesn't elevate the entire user, it only elevates the application.
Right, which is why I referred to the "app in question".
For the entire length of execution of that application, the application will run elevated. For instance, Visual Studio.NET is an application that pretty much always needs to be run as admin.
Speaking of spectacularly bad ideas...
If you cached the elevated credentials authorization for "X" minutes, or whatever, you would be giving a free pass to any malware that happened to be trying to do something bad.
The elevated permissions would apply per-process. The app would request permission, do something, and drop the permissions. If it requested it again within the timeout, it'd get the permission again without a prompt.
No, it doesn't. By default, all users on the system can read files in c:\Program Files.
The hell it doesn't.
It does throw up a UAC, but I've already explained why the "caching period" is a bad idea.
You misunderstood the idea of the timeout. It would indeed be inane to elevate the user for the entire duration. Note that 'sudo' doesn't do that.
The better solution is what OS X does: extend "sudo" to the GUI. The first time the app needs escalated privileges, prompt for the user's password. Then, cache those privileges for a reasonable amount of time and don't prompt. Unless the app in question is compromised in that interval, it doesn't matter.
The problem with UAC is that it fails to separate the two orthogonal issues of sanity-checking the user's behavior, and maintaining system security. Consider how "Program Files" is handled. Browsing into "Program Files" throws up a UAC alert. It shouldn't do that --- "Program Files" is readable to everyone. Writing to "Program Files" should throw up a UAC alert, but only the first time in the caching period. The question at that point isn't "Do you really want to modify this directory" (of course I do!), but rather "Do you want to give Explorer.exe permission to modify this directory". When you follow the first train of thought, you end up with prompting the user each time, because obviously each copy requires a separate sanity-check. If you follow the second train of thought, you see that the caching mechanism is just fine, since if Explorer.exe was authorized 30 seconds ago, it's unlikely it was compromised since then, and should retain that authorization.
The "next, next, next" type install has a lot of pitfalls. First of all is the whole idea of an installer program, which is certainly not an intuitive concept. It's one thing to start a program to browse for other programs (Synaptic), or to have a program just be a bundle you can download from a website, but to have a program that you download actually be a proxy to install the program you really want? Not obvious. Then there is the fact that most inexperienced people see all the options in the installer and freak. For most software, none of those options matter! That's why you people derisively call them "Next, next, next" installers, because you can just mash the "Next" button the whole time and get the correct result. Aside from a few rare, big applications (where choosing components is important), a simple "install the fucking app already" button is all you really need.
Most of those prompts were redundant, either because they enforce things guaranteed by the underlying file permissions, or because the authorization could've been cached.
Vista:Are you sure you want to run this program?
Of course! It's got +X set!
Vista: are you sure you want to run this? it's a program, you know.
Ditto.
Vista: This is from AMD. Do you trust AMD?
Redundant. If I didn't trust them, I wouldn't have set +X.
Vista: Are you sure you want to go there?
Since Program Files shouldn't be world writable, this should prompt you for the administrator password. This authoriation should then be cached for Explorer.exe.
Vista: Are you sure you want to overwrite this file?
I'll let this slide, because even 'cp' prompts for that.
Vista:A program wants to write to the Program Files folder. Is this ok?
Should've grabbed cached authorization for Explorer.exe. Unless Explorer.exe was compromised in the 30 seconds between this action and the previous one, no security is lost here.
Vista:You are trying to copy from a network share to the program files folder. This isn't allowed. Hit ok.
That's just idiotic.
Are you sure you want to overwrite this file?
Again, I'd let it slide depending on preference.
Vista:A program wants to write to the Program Files folder. Is this ok?
Cached authorization again.
It's really not that hard. UNIX/sudo got this right god knows how long ago. Apple did the right thing and just copied the sudo mechanism wholesale. Microsoft should to.
How many grandmas can figure out a Windows "next, next, next" type install?
Especially since you can use the "Add/Remove" programs applet in Ubuntu to (gasp!) actually add programs (unlike in Windows), your rant is kind of hollow.
No, because the native Americans never founded a country here.
Or maybe they're taking the money saved on the "slower" RAM (that isn't any slower in actual use) and spending it on better motherboards or PSUs ?
Which one do you really thing is more likely? Honestly? If they're trying to save a few bucks on something like RAM, which is a spec they can advertise, what do you think they're doing with the stuff they can't advertise. I've been around too many Dells in the last few years to give them the benefit of the doubt.
Seriously, it's a non-issue. Dell still have options for 533Mhz DDR2 on machines like their PowerEdge 2950s, which aren't quite solid and reliable servers.
The 2950 uses server-grade, high-capacity FB-DIMMs, which are a whole different kettle of fish from what they're putting in the E520! When you see a lower speed rating on a server DIMM that costs $3000, you assume conservativeness, not cost-cutting. I have a hard time believing the 533 MHz DIMMs in the E520 are otherwise high-quality parts that are just conservatively rated. More likely they are low-quality parts sold at a discount to get rid of the inventory of 533 MHz chips.
Apple do the case design and maybe a bit of internal layout to help get things into the case they want, but now they've switched to Intel pretty much everything is off-the-shelf and built in the same factories pumping out regular old PC components - because they *are* regular old PC components.
Sure, Mac's use a lot of the same parts as PCs. However, Apple still has control over parts selection, and still custom-designs their motherboards, power supplies, cases, and cooling systems. In my experience with building PCs, these are the reliability bottlenecks. I rarely see PCs where the Intel CPU or chipset craps out. On the other hand, I quite often see PCs where the motherboard is a cheap 3-layer piece of shit, or the "500W" power supply can't reliably deliver half that, or where the case is made of cheap plastic and has flimsy bits that break off easily. You're not going to find that stuff in an Apple machine. Of course, you can find it in the PC world too (think ThinkPads, or building it yourself), but then you're going to have to be prepared to spend as much money as you would on an Apple machine.
Added to that, even if they did do the whole design top to bottom, that still wouldn't impact build quality. You can have the best architect in the world, but if the builders and incompetent and the cement bad, the building is still going to fall down.
The build quality is largely a factor of how much you're willing to spend. You can easily have 0.01" tolerance on a spec, instead of 0.1" tolerance, even with the same builder, it's just going to cost you more money. Dell's suppliers in Taiwan are probably easily capable of building a machine with the same tolerances as an Apple, but it seems that since about 2000 or so, Dell isn't willing to pay for that.
iMacs (and Mac Minis and MacBooks) are built to a budget. Mac Pros (and the other "pro" machines aren't (well, are less so), and carry a suitable price premium to compensate (and they _are_ very nice inside - well, the MBPs are a bit of a mess...). You can get the same thing from Dell, as well, if you want to spring for a Precision workstation or higher-end Latitude (vs Inspiron) notebook (or go IBM/Lenovo).
Every machine is build towards a target price. However, Apple seems to be less willing to trade user experience for that target price than does Dell. Even the MacBook has the same close-flush design that causes Apple's cases to cost more to build. Even the iMac uses high-quality acrylics and hard plastics that cost more than the softer stuff you see in most Dells. Apple could've saved quite a bit of money on the iMac by making it an inch thicker, specifying a more powerful (and hence louder) fan, and using desktop parts instead of laptop ones. They didn't because they didn't want to trade off the user experience.
To be fair, Apple isn't the only one who builds machines like this. Lenovo's Th
see, this is the problem with a mac fan, you think that someone actually wants the parts just how apple give them. I know its hard for you, but try to imagine this isn't my first computer and most of those gadgets I already own.
If you don't want the parts, then fine, don't buy the machine. But don't say you can get a similar machine for a lot less money (which is what you did!) If you want to compare price, then spec out a roughly comparable machine, and compare the price. If you want to argue about flexibility, then argue that instead. I'm perfectly happy to admit that Apple's canned configurations are nowhere near as flexible as building something yourself, that's really a no-brainer.
now, even after buying a second monitor, I have over 600 bucks left over. I'm comparing with the tower now, which is at least 2500 bucks w/o monitor, much less dual monitors. I'm quote happy with my monitor so I'd get the same one for a mac tower if I had bought one. Just the box, I'm at half the price, far more powerful than any iMac I could have gotten, and with my use of the computer, close enough to a mac tower at the time.
The Mac Pros are just about six months old. They're all quad core (2 x Xeon), and always have been. They're completely out of the league of your machine. In addition to the pair of Xeons, they've got a server-grade Intel 5000X motherboard that'll run you $400-$500 by itself. They've got quad-channel FB-DIMM memory and 8 memory expansion slots. And the parts alone will cost you about as much as a Mac Pro, even neglecting labor, warranty, etc.
Firstly, 533Mhz DDR2 isn't "last-gen tech". DDR1 is "last gen tech". There are numerous machines out there still only using 533Mhz RAM - and most of the ones using 667Mhz RAM derive no benefit except in a handful of synthetic benchmarks.
The issue with DDR2-533 is that it's kind of a bastard-stepchild speed grade. It had a run of only about a year, after being introduced with the earliest Intel DDR2 platforms and supplanted by DDR2-667 very shortly afterward. DDR2-667 has been common now for almost two years, and as you said the price difference is minimal. Since all Core 2 chipsets support DDR2-667, the only reason to ship the older memory in a machine would be to save a few extra cents. It's that sort of scrimping that worries me. After all, if they don't want to pony up a couple of extra bucks for RAM, who knows how much they're skimping on the PSU or motherboard.
The iMac is assembled to a budget - to save "every last penny" - and with the cheapest parts available to fill its requirements (which happen to include a particularly small form factor, requiring some low-power parts). Don't kid yourself otherwise. Like I said, iMacs roll off the same assembly lines PCs like the E520 do. Heck, MacBook Pros are just rebadged Asus laptops (or certainly used to be).
This is actually very much incorrect. Yes, the same Taiwanese manufacturers (like Asus) actually build the machines, but Apple is the one that does the design (no Apple laptop is a re-badged anything). Moreover, it's well known in the industry that Apple's tolerances are a lot tighter than everyone else's. Take their laptops for example. The MacBooks has a case design that is really hard to build. When closed, you can run your finger along the seam between the lid and the body, and it should be tight and flush. Even a slight misalignment results in a noticeable defect. As a result, they have to specify stricter tolerances on the build, which costs money. Almost every other laptop design includes some sort of rounded lip, which can hide substantial misalignment without the user noticing, and can thus be manufactured more cheaply. Also, take a look inside a PowerMac or Mac Pro sometime. Apple doesn't skimp on things like the PSU, motherboard, or even case fans. A lot of manufacturers do, because a great PSU is not something they can advertise in the spec sheet. Apple's PowerMac PSUs run from 700W to 1000W (real, not like the $30 "500W" PSUs you see on NewEgg), and they're heavy and solid like a PSU should be. The PowerMac (and probably Mac Pro)'s fans are a custom variant of a Delta model that runs $10-$12 on the internet. You're not going to find a $10+ case fan in that E520.
Note that Dell used to build machines like this too. I've got a Dimension D300 from 1998, and its probably as heavy as a modern E-series machine despite being substantially smaller. It uses more expensive hard plastics (as opposed to the cheaper soft plastics you see on modern Dells), and an Intel Seattle motherboard modified for better power delivery. The damn thing is getting close to 10 years old, and is still running flawlessly. A couple of years ago it was sitting in a closest acting as a server while the room was remodeled. I realized after a couple of months that it had been inhaling all the dust from the remodeling for all that time, yet it never even so much as crashed.
Now, Apple certainly isn't as premium as you can get. I've spent $20+ on a CPU fan because that has a drastic effect on the noise profile of the machine. I've spent $130 on an Antec P180 case because of the noise-absorbing aluminum/plastic/aluminum sandwich paneling. These are a cut above even what you find in an Apple machine. However, I've dug around my fair share of cheaper PCs, and almost all of them cut corners on these sorts of parts.
You seem to have that arse-about-face. Gaming - on of the few things the average person will do that actually stretches a modern computer - is something where you most certainly can *not* get away with "a really shitty machin