Now, you buy Office and you use it. Forever. I'm pretty sure I'm not the only one who still has an Office97 running somewhere. Why upgrade?
Because you're getting documents created with a more recent version of Office and you can't edit them without mucking up the formatting, so your customers/vendors/employer/etc convince you to upgrade.
If you get a Pay As You Go phone, that's your phone. The cellular company basically gives away the phone to sell the communication service, and the PAYG phone is rarely a "full featured" phone... it's got no value outside the cellular system.
That is, you're not buying the phone. You're buying the service.
This is just hire-purchase, rent-to-own. The 800 hours of service aren't optional, they're just a different way to pay. The user has to pay them all - they're on the hook for 800 hours, even if they have more control over how fast they pay it off. And the scheme really depends on people paying off the PCs quickly: if it turns out that people are taking too long to pay the financial institutions are going to change the terms to a more conventional HP scheme. They'll have to, THEY'RE on the hook too.
The best antivirus protection is not to be exposed to viruses.
Like the guy in the first part of Egan's "Distress" who was converting himself to use a different set of amino acids in his DNA/RNA to render himself immune to all viruses (except for rogue versions of the ones used to implement the conversion, which seemed to be a rather obvious hole in his clever scheme to me), the best protection is to be a different species, genus, phylum, or kingdom than the infected population.
Failing that, I've fallen back on "Don't use Internet Explorer or any other application that uses the Microsoft HTML control, damnit". That at least turns the antivirus clock back to 1995 or so, when the biggest exposure was still shareware and local network exploits. I can live with that.
I'm not quite sure why you think the Mini is targeting a much different market to the MacBook.
Did you catch the "reluctantly" bit in there?
I'm talking about Apple apologists who beable on slashdot about how the crappy video in the Mini is OK because it's "targeted for a different market".
Not to mention, everything running OS X is running 3D, not 2D.
There is absolutely no T&L or any other advanced 3d required by the OSX user interface. It's all throwing around scaled 2d bitmaps in a 2d plane with fake shadows done by layering 2d shadow bitmaps at the edge of the window. It's no more "3d" than Windows NT 4.0 or X11 with Motif... it's a 3d look using 2d bitmap operations.
None of its contemporaries are.
I'm seeing a number of sub-1000 notebooks with Radeon chipsets.
Games are - in the realm of non-professional software - the most hardware-intensive software the average person will ever use. Why on earth would anyone think a low-end general-purpose computer would be capable of running such high-end, specific software well?
I didn't say "well", I said "better than barely adequately".
And I said "better than barely adequately" rather than "even barely adequately" because the post I was responding to described it as being barely adequate. I wouldn't use that phrase to describe the existing iBook's performance... I'd have called it "embarassingly bad" or even "unacceptable"... but if that's what you call "barely adequate" then I want more than what you consider "barely adequate".
The GPU on the Macbook Pro isn't a "gamer chip" either, but it's adequate.
I accepted the horrible performance of the iBook and Powerbook because they had an excuse... the low speed memory bus on the G4. Once that bottleneck was gone, they should be able to do at least do as well as other entry level computers with low-end-but-acceptable chipsets like the nVidia go5200.
So now instead of being crippled by the 166 MHz bus, they're crippled by the Intel GMA950. They're still crippled.
It's like buying a Honda Civic and then complaining that it sucks on the racetrack.
No, it's like buying a Honda Civic and discovering it can't accelerate to freeway speed by the end of the on-ramp.
If you want to play games, either buy a computer meant for playing games, or buy one high-end enough that it can play games as a side effect.
Most of the OpenGL apps I'm interested in aren't actually games, but they run acceptably fast on any machine with any GPU that actually implements a full set of 3d shaders. That is, anything that doesn't actually suck at games. I was hoping that the higher speed bus on the Intel Core would mean Apple was finally going to ship machines that were comparable to the better entry level (under $500) PCs or (under $1000) laptops. What a fool I was. Apple still thinks that anyone who isn't interested in forking over $2000 at the drop of a hat doesn't matter.
The situation is different for games, but this is a machine that simply isn't meant to be used for gaming.
I was, reluctantly, willing to accept that line for the Mini in the context of the home entertainment center... where good 2d performance was the goal.
But a consumer laptop is a personal computer, and I bloody well DO expect it to be "more than barely adequate" for playing games. Especially when the excuse given for the previous generation being barely adequate was the anemic memory subsystem on the CPU, and the whole point of this generation was to fix that.
ESPECIALLY given how cheap an adequate video card is.
Intel must REALLY be taking Apple to the cleaners.
I think you're assuming a half-assed implementation of a sandbox, e.g. one that allows the sandboxed app to make arbitrary system calls. A proper sandbox will ensure that the software running within the sandbox has no way to access anything except what it absolutely needs to access in order to display the file.
That'd be great, except that any software you actually want to run in a sandbox is going to have to access an awful lot of crap to just start up and run, let alone display a file. So, yeh, unless you want to spend an enormous amount of time and energy trying to out-think the idiot application developers you're going to have to fire up a virtual machine for EVERY instance of the application, or your sandbox is going to be default-allowed unti it's leakier than a seive, or it's going to be so complex it's more likely to have a bug than any of the apps running under it.
OR, you write a sandbox that's only going to run applications designed to cooperate with the sandbox. But that's what the web browser is, in the first place. If the developer of the application hasn't bothered to create a standalone secure viewer, let alone a Java plugin that cooperates wit the java sandbox, what makes you think he's going to cooperate with your new restricted its-not-java-but-it-might-as-well-be sandbox?
Baby steps. Let's get the part that we absolutely need (secure apps, and browsers that ONLY use secure apps) and worry about elaborations later.
But if you are like me and like to use it like a book, to carry it with one hand with no bag when you walk short distances, to being sit comfortably on a couch with the laptop laid on ONE lap (because it's so small), etc, then 5 lbs vs. 2 lbs makes a lot of differences.
I use a Palm for that kind of thing. I tried it with a Toshiba Libretto - which was an amazingly nice laptop for its time - but once I got my first Palm I hardly ever carried a laptop anywhere.
Pity Palm's forgotten the point of a handheld in recent years. I got several devices, even tried a Pocket PC, but I stopped upgrading a few years back. Palm Vx, Visor Deluxe, iPaq, Visor Prism, Jornada, Clie... the last Palm I bought was a Sony SJ-22 and I've yet to see a handheld that I really feel would be an upgrade from it. It's a bit shopworn now and overdue for replacement and I'm probably going to end up looking for a refurb SJ22 to replace it.
I doubt if I'd have any trouble learning that one finger gives me a left click, while putting down the second finger gives me a right click.
I typically click with *no* fingers down on the touchpad itself unless I'm dragging - I have too much trouble with touches turning into drags otherwise.
As for your thumb or your little finger, they could simply put the "right button" above the trackpad. I've used trackpads like this, with the "left button" below, and the "right" and "middle" buttons above.
How about a hack that turns right-control into right-click? Not click-right-control, just plain right-control. Anyone have something like that?
[Re: getting rid of the XPI install option] Unlikely to happen, I would've thought.
Probably so, given the level they've gone to to convince themselves it's safe.
It should still be done.
At the very least it should be easy and obvious to install XPI from the local disk. Last time I tried this the easiest way seemed to be to install an extension for it!
Once you allow local code execution, even in an OS-level sandbox, you've halfway lost the battle. Now the attacker has the ability to make arbitrary system calls, access any resources that the sandboxed application would need, and has in general a MUCH stronger place to work from. Not only that, but an exploit in the sandbox will be much harder to fix since any application that was legitimately using whatever feature the exploiter used will need to be upgraded.
So as an additional line of defence a sandbox is useful, but it doesn't change the absolute requirement that you only use "safe applications" to view untrusted content. And it's a lot easier to fix a "safe application" than either fixing a sandbox or fixing a hole in a general purpose application, since people are more willing to accept limitations required by security in a "viewer".
Then it won't matter if the program is "safe" or not, because even if the program gets compromised, the malware can never leave the process's sandbox.
Unfortunately, it does matter, because the security of a sandbox is compromised by every legitimate access requirement that an application might have. That's why Microsoft refused to sandbox ActiveX... they saw the restrictions a sandbox would impose on an application as too great a cost.
Your G4's got a Radeon 9600, and I would be surprised if it doesn't match the performance of the Macbook 1.83, if not better it. Why don't you give it a shot and post the results here?
Test number two: try playing the high-definition (1080i) trailers on Apple's website.
The GMA chipset is pretty good for 2d acceleration, it's 3d it sucks at. That's why people excused it in the new mini since it was "targeted for home entertainment centers".
They don't have that excuse in the MacBook.
Intel must really be soaking them for their Core Duo chips if they have to cripple the boxes elsewhere to keep the price down.
So what's worse? Integrated graphics or an underclocked Radeon X1600?
What's worse? Integrated graphics.
The Core Duo mini using software OpenGL plus the GPU barely edges out the 166 MHz bus G4 + Radeon 9200 in the original mini in some benchmarks. It's up to snuff with modern Radeons in others, but *damn*... having to use a dual core CPU to make sure you sweep the grotty old Radeon 9200 should be terminally embarassing for Apple.
That's got the same problem as the stupid "Mighty Mouse"... it's too easy to accidentally register the wrong button (I can't reliably right-click with the MM, I have to stop and deliberately lift my index finger), plus you can't chord for the middle button in X11 apps.
Doex anybody really like two-button trackpads? I feel like I'm flirting with carpal tunnel every time I twist my hand into that awkward position needed to reach the right button with my thumb
Don't do that then. Hit it with your little finger or your ring finger.
You get support for the full 3D pipeline in hardware, including shaders.
No you don't. A lot is emulated on the CPU, that's why the Radeon 9200 on the Mac mini actually beats the Core Solo mini for hardware T&L, and ties with the Core Duo.
I've used the Intel chipset under Windows XP, and the performance with 3d apps was miserable... my old celeron 1.7 with an nVidia 5600 was smoother than a 3 GHz P4 with Integrated Graphics. It was still faster than my older Radeon 9200 for most stuff (but not always noticably) but *damn*.
The only place it really kicked butt was video playback.
Now, my Celeron is probably faster than the old Powerbook simply because I have a MUCH faster memory bus... so it'll probably beat the old iBook most of the time. And that's Windows XP, but I've run the Mac version of some of the same software on my Mini and it wasn't that far behind the 3 GHz P4 with the Intel graphics.
And... I've seen some game benchmarks where the iBook beat the Macbook or the Intel mini even with universal binaries. Now, I suppose on the mini they could say "this is an entertainment center, it's not a general consumer machine"... but the Macbook *is* targeting the general consumer market, and if the new Macbook's slower for games than the iBook was that's just not acceptable.
In my 20 years of system administration I have often had people come to me and say "Peter, I just clicked the wrong button and my computer's acting funny." I've less often had people say "Peter, I downloaded a file to the desktop and opened it and my computer's acting funny." I've had several people say "Peter, I just clicked the wrong button AGAIN and I think I'm infected."
I've never had the same person come to me twice with "I've downloaded and opened a file and I'm infected." Give people even a small breathing space to think about what they're doing, without that reflex "gotta push a button" effect, and social social engineering is MUCH harder.
So...
You can solve this for most people simply by not including a mechanism for running untrusted content. Don't pop up a dialog box asking "What do you want to do with this application you just downloaded? (Open) (Show) (Ignore)". Don't even ask "The file you just asked to open is an appliaction? (Infect Me) (Cancel)". Just don't put the user in the position of deciding, right then, what to do with the file. Ever.
Firefox: get rid of the XPI install-from-web stuff. Let the user download the XPI and open it explicitly.
Apple: Dont' "open safe files after downloading"... there are no "safe files".
Microsoft: get rid of ActiveX and security zones and for god's sake don't try and make.NET-in-the-browser into the next Active Desktop disaster.
All of the above: If it's a file you've got a safe application for... a *safe application*, not a *safe file*... open it explicitly IN THAT APPLICATION. Don't go "this is a ZIP file so I'll open it in whatever random program the user has for opening archives". Keep a database of safe programs to use on untrusted content like you keep a database of plugins people have explicitly installed. This would resolve SO MANY security issues... damnit.
(don't treat archives as "safe files", but that's another rant)
Those Cinebench results are so far off the other results I've seen for the Intel chipset that I'm certain they weren't significantly dependent on the GPU... they track the CPU speed difference too closely. Software that actually uses the GPU for 3d gives a very different result.
I use "vi" on real plaintext documents myself, so I'm unlikely to hit this, but anything you know that might keep the fuckup fairy away is all to the good.:)
Slashdot Mirror
About the brazilian tests, maybe this is a big secret program...
Only 1000 PCs? That's barely test-marketing, you can bet they cherry-picked their prospects.
Now, you buy Office and you use it. Forever. I'm pretty sure I'm not the only one who still has an Office97 running somewhere. Why upgrade?
Because you're getting documents created with a more recent version of Office and you can't edit them without mucking up the formatting, so your customers/vendors/employer/etc convince you to upgrade.
This is classic "creative incompetance".
If you get a Pay As You Go phone, that's your phone. The cellular company basically gives away the phone to sell the communication service, and the PAYG phone is rarely a "full featured" phone... it's got no value outside the cellular system.
That is, you're not buying the phone. You're buying the service.
This is just hire-purchase, rent-to-own. The 800 hours of service aren't optional, they're just a different way to pay. The user has to pay them all - they're on the hook for 800 hours, even if they have more control over how fast they pay it off. And the scheme really depends on people paying off the PCs quickly: if it turns out that people are taking too long to pay the financial institutions are going to change the terms to a more conventional HP scheme. They'll have to, THEY'RE on the hook too.
The best antivirus protection is not to be exposed to viruses.
Like the guy in the first part of Egan's "Distress" who was converting himself to use a different set of amino acids in his DNA/RNA to render himself immune to all viruses (except for rogue versions of the ones used to implement the conversion, which seemed to be a rather obvious hole in his clever scheme to me), the best protection is to be a different species, genus, phylum, or kingdom than the infected population.
Failing that, I've fallen back on "Don't use Internet Explorer or any other application that uses the Microsoft HTML control, damnit". That at least turns the antivirus clock back to 1995 or so, when the biggest exposure was still shareware and local network exploits. I can live with that.
And of course, no modern anti-apple troll should be complete without reference to Apple's 'betrayel' of PA semi
Don't forget MPC8641D.
You know you can run Linux on a handheld?
I'm not quite sure why you think the Mini is targeting a much different market to the MacBook.
Did you catch the "reluctantly" bit in there?
I'm talking about Apple apologists who beable on slashdot about how the crappy video in the Mini is OK because it's "targeted for a different market".
Not to mention, everything running OS X is running 3D, not 2D.
There is absolutely no T&L or any other advanced 3d required by the OSX user interface. It's all throwing around scaled 2d bitmaps in a 2d plane with fake shadows done by layering 2d shadow bitmaps at the edge of the window. It's no more "3d" than Windows NT 4.0 or X11 with Motif... it's a 3d look using 2d bitmap operations.
None of its contemporaries are.
I'm seeing a number of sub-1000 notebooks with Radeon chipsets.
Games are - in the realm of non-professional software - the most hardware-intensive software the average person will ever use. Why on earth would anyone think a low-end general-purpose computer would be capable of running such high-end, specific software well?
I didn't say "well", I said "better than barely adequately".
And I said "better than barely adequately" rather than "even barely adequately" because the post I was responding to described it as being barely adequate. I wouldn't use that phrase to describe the existing iBook's performance... I'd have called it "embarassingly bad" or even "unacceptable"... but if that's what you call "barely adequate" then I want more than what you consider "barely adequate".
The GPU on the Macbook Pro isn't a "gamer chip" either, but it's adequate.
I accepted the horrible performance of the iBook and Powerbook because they had an excuse... the low speed memory bus on the G4. Once that bottleneck was gone, they should be able to do at least do as well as other entry level computers with low-end-but-acceptable chipsets like the nVidia go5200.
So now instead of being crippled by the 166 MHz bus, they're crippled by the Intel GMA950. They're still crippled.
It's like buying a Honda Civic and then complaining that it sucks on the racetrack.
No, it's like buying a Honda Civic and discovering it can't accelerate to freeway speed by the end of the on-ramp.
If you want to play games, either buy a computer meant for playing games, or buy one high-end enough that it can play games as a side effect.
Most of the OpenGL apps I'm interested in aren't actually games, but they run acceptably fast on any machine with any GPU that actually implements a full set of 3d shaders. That is, anything that doesn't actually suck at games. I was hoping that the higher speed bus on the Intel Core would mean Apple was finally going to ship machines that were comparable to the better entry level (under $500) PCs or (under $1000) laptops. What a fool I was. Apple still thinks that anyone who isn't interested in forking over $2000 at the drop of a hat doesn't matter.
The situation is different for games, but this is a machine that simply isn't meant to be used for gaming.
I was, reluctantly, willing to accept that line for the Mini in the context of the home entertainment center... where good 2d performance was the goal.
But a consumer laptop is a personal computer, and I bloody well DO expect it to be "more than barely adequate" for playing games. Especially when the excuse given for the previous generation being barely adequate was the anemic memory subsystem on the CPU, and the whole point of this generation was to fix that.
ESPECIALLY given how cheap an adequate video card is.
Intel must REALLY be taking Apple to the cleaners.
I think you're assuming a half-assed implementation of a sandbox, e.g. one that allows the sandboxed app to make arbitrary system calls. A proper sandbox will ensure that the software running within the sandbox has no way to access anything except what it absolutely needs to access in order to display the file.
That'd be great, except that any software you actually want to run in a sandbox is going to have to access an awful lot of crap to just start up and run, let alone display a file. So, yeh, unless you want to spend an enormous amount of time and energy trying to out-think the idiot application developers you're going to have to fire up a virtual machine for EVERY instance of the application, or your sandbox is going to be default-allowed unti it's leakier than a seive, or it's going to be so complex it's more likely to have a bug than any of the apps running under it.
OR, you write a sandbox that's only going to run applications designed to cooperate with the sandbox. But that's what the web browser is, in the first place. If the developer of the application hasn't bothered to create a standalone secure viewer, let alone a Java plugin that cooperates wit the java sandbox, what makes you think he's going to cooperate with your new restricted its-not-java-but-it-might-as-well-be sandbox?
Baby steps. Let's get the part that we absolutely need (secure apps, and browsers that ONLY use secure apps) and worry about elaborations later.
But if you are like me and like to use it like a book, to carry it with one hand with no bag when you walk short distances, to being sit comfortably on a couch with the laptop laid on ONE lap (because it's so small), etc, then 5 lbs vs. 2 lbs makes a lot of differences.
I use a Palm for that kind of thing. I tried it with a Toshiba Libretto - which was an amazingly nice laptop for its time - but once I got my first Palm I hardly ever carried a laptop anywhere.
Pity Palm's forgotten the point of a handheld in recent years. I got several devices, even tried a Pocket PC, but I stopped upgrading a few years back. Palm Vx, Visor Deluxe, iPaq, Visor Prism, Jornada, Clie... the last Palm I bought was a Sony SJ-22 and I've yet to see a handheld that I really feel would be an upgrade from it. It's a bit shopworn now and overdue for replacement and I'm probably going to end up looking for a refurb SJ22 to replace it.
I doubt if I'd have any trouble learning that one finger gives me a left click, while putting down the second finger gives me a right click.
I typically click with *no* fingers down on the touchpad itself unless I'm dragging - I have too much trouble with touches turning into drags otherwise.
As for your thumb or your little finger, they could simply put the "right button" above the trackpad. I've used trackpads like this, with the "left button" below, and the "right" and "middle" buttons above.
How about a hack that turns right-control into right-click? Not click-right-control, just plain right-control. Anyone have something like that?
[Re: getting rid of the XPI install option] Unlikely to happen, I would've thought.
Probably so, given the level they've gone to to convince themselves it's safe.
It should still be done.
At the very least it should be easy and obvious to install XPI from the local disk. Last time I tried this the easiest way seemed to be to install an extension for it!
Once you allow local code execution, even in an OS-level sandbox, you've halfway lost the battle. Now the attacker has the ability to make arbitrary system calls, access any resources that the sandboxed application would need, and has in general a MUCH stronger place to work from. Not only that, but an exploit in the sandbox will be much harder to fix since any application that was legitimately using whatever feature the exploiter used will need to be upgraded.
So as an additional line of defence a sandbox is useful, but it doesn't change the absolute requirement that you only use "safe applications" to view untrusted content. And it's a lot easier to fix a "safe application" than either fixing a sandbox or fixing a hole in a general purpose application, since people are more willing to accept limitations required by security in a "viewer".
Then it won't matter if the program is "safe" or not, because even if the program gets compromised, the malware can never leave the process's sandbox.
Unfortunately, it does matter, because the security of a sandbox is compromised by every legitimate access requirement that an application might have. That's why Microsoft refused to sandbox ActiveX... they saw the restrictions a sandbox would impose on an application as too great a cost.
UT2004 at Macworld:
iBook G4 1.42, nVidia go5200 - 14.1 FPS
Macbook 1.83, GMA950 - 17.8 FPS
Powerbook 1.67, Mobility Radeon 9700 - 21.4 FPS
Macbook Pro 2.16, Mobility Radeon X1600 - 63.1 FPS
Your G4's got a Radeon 9600, and I would be surprised if it doesn't match the performance of the Macbook 1.83, if not better it. Why don't you give it a shot and post the results here?
In MacWorld's benchmarks with real-world OpenGL (UT2004), the MacBook Pro, with real video, delivered three times the framerate of the MacBook.
Nice one. Notice that the Macbook's frame rate about split the difference between the iBook and the Powerbook?
iBook - 14.1
Macbook - 17.8
Powerbook - 21.4
Mcbook Pro - 63.1
So the Core Duo and the GMA graphics together were somewhere between the performance of the nVidia go5200 and the Radeon 9600.
Test number two: try playing the high-definition (1080i) trailers on Apple's website.
The GMA chipset is pretty good for 2d acceleration, it's 3d it sucks at. That's why people excused it in the new mini since it was "targeted for home entertainment centers".
They don't have that excuse in the MacBook.
Intel must really be soaking them for their Core Duo chips if they have to cripple the boxes elsewhere to keep the price down.
When they say that it's "4-5x faster" but you're using all that extra speed to cover up for the crummy GPU... that's just pitiful.
So what's worse? Integrated graphics or an underclocked Radeon X1600?
What's worse? Integrated graphics.
The Core Duo mini using software OpenGL plus the GPU barely edges out the 166 MHz bus G4 + Radeon 9200 in the original mini in some benchmarks. It's up to snuff with modern Radeons in others, but *damn*... having to use a dual core CPU to make sure you sweep the grotty old Radeon 9200 should be terminally embarassing for Apple.
That's got the same problem as the stupid "Mighty Mouse" ... it's too easy to accidentally register the wrong button (I can't reliably right-click with the MM, I have to stop and deliberately lift my index finger), plus you can't chord for the middle button in X11 apps.
Doex anybody really like two-button trackpads? I feel like I'm flirting with carpal tunnel every time I twist my hand into that awkward position needed to reach the right button with my thumb
Don't do that then. Hit it with your little finger or your ring finger.
You get support for the full 3D pipeline in hardware, including shaders.
No you don't. A lot is emulated on the CPU, that's why the Radeon 9200 on the Mac mini actually beats the Core Solo mini for hardware T&L, and ties with the Core Duo.
I've used the Intel chipset under Windows XP, and the performance with 3d apps was miserable... my old celeron 1.7 with an nVidia 5600 was smoother than a 3 GHz P4 with Integrated Graphics. It was still faster than my older Radeon 9200 for most stuff (but not always noticably) but *damn*.
... so it'll probably beat the old iBook most of the time. And that's Windows XP, but I've run the Mac version of some of the same software on my Mini and it wasn't that far behind the 3 GHz P4 with the Intel graphics.
The only place it really kicked butt was video playback.
Now, my Celeron is probably faster than the old Powerbook simply because I have a MUCH faster memory bus
And... I've seen some game benchmarks where the iBook beat the Macbook or the Intel mini even with universal binaries. Now, I suppose on the mini they could say "this is an entertainment center, it's not a general consumer machine"... but the Macbook *is* targeting the general consumer market, and if the new Macbook's slower for games than the iBook was that's just not acceptable.
In my 20 years of system administration I have often had people come to me and say "Peter, I just clicked the wrong button and my computer's acting funny." I've less often had people say "Peter, I downloaded a file to the desktop and opened it and my computer's acting funny." I've had several people say "Peter, I just clicked the wrong button AGAIN and I think I'm infected."
.NET-in-the-browser into the next Active Desktop disaster.
I've never had the same person come to me twice with "I've downloaded and opened a file and I'm infected." Give people even a small breathing space to think about what they're doing, without that reflex "gotta push a button" effect, and social social engineering is MUCH harder.
So...
You can solve this for most people simply by not including a mechanism for running untrusted content. Don't pop up a dialog box asking "What do you want to do with this application you just downloaded? (Open) (Show) (Ignore)". Don't even ask "The file you just asked to open is an appliaction? (Infect Me) (Cancel)". Just don't put the user in the position of deciding, right then, what to do with the file. Ever.
Firefox: get rid of the XPI install-from-web stuff. Let the user download the XPI and open it explicitly.
Apple: Dont' "open safe files after downloading"... there are no "safe files".
Microsoft: get rid of ActiveX and security zones and for god's sake don't try and make
All of the above: If it's a file you've got a safe application for... a *safe application*, not a *safe file*... open it explicitly IN THAT APPLICATION. Don't go "this is a ZIP file so I'll open it in whatever random program the user has for opening archives". Keep a database of safe programs to use on untrusted content like you keep a database of plugins people have explicitly installed. This would resolve SO MANY security issues... damnit.
(don't treat archives as "safe files", but that's another rant)
(in fact there's a lot of ranting I could add here...)
Those Cinebench results are so far off the other results I've seen for the Intel chipset that I'm certain they weren't significantly dependent on the GPU... they track the CPU speed difference too closely. Software that actually uses the GPU for 3d gives a very different result.
Didn't IBM require that Lenovo put US people in charge of their laptop group as a condition of sale?
Thanks for the warning.
:)
I use "vi" on real plaintext documents myself, so I'm unlikely to hit this, but anything you know that might keep the fuckup fairy away is all to the good.