Configuration files which change by user shouldn't be stored in Program Files, they should be in the user's profile directory.
I think the point is that configuration files that DON'T change by the user are still (by default) in the console user's profile directory.
Suppose I have something like IIS for the server, and I'm an admin user, and IIS is configured by a text file. I launch some kind of IIS configuration editor, which opens "C:\Program Files\IIS\foobaz.cfg" for edits. However, the file really being edited is "C:\Users\me\...\foobaz.cfg".
Well, here's the problem, as I see it. The command line is an expressive interface, but largely limited to character-cell based interfaces. I just think we need to get out of that mode of thought. If you can take the command line out of simple character cells, that would make it even *more* expressive.
(Disclaimer: I wrote a console-based terminal emulator to bring BBS-era convenience to the Linux console. Works pretty well, and can be very convenient at times.)
I think the issue is that the character-cell metaphor is a "good enough" interface for mouse-less work, especially with ncurses encapsulating 30 years of text terminals development. You can still add mouse and go (T)IMP too: see the OSS port of Turbo Vision. This metaphor also works over the simplest available two-way links ala ssh/telnet/serial port/etc. Once you break out of the character-cell metaphor, you've got the full X11 GUI that works over network too. The question is why would you seek any kind of intermediate ground when both metaphors already work OK?
My personal opinion is that the next great interface revolution won't occur until either a) voice commands become the norm, or b) we can project a HUD-like screen over our real sight (total immersion / augmented reality). I suspect B will happen first, and it will probably come in a form like the Nintendo WII where multiple 3D pointers are used.
Strange. You're for both a flat tax and the Basic Income Guarantee. I've never seen that. Does your libertarian right hand want to cut off your socialist left hand?
So what exactly IS "relevant" to this discussion? Only consumer desktop machines that might play HD video across HDMI from a Blu-ray source?
My embedded system is a bit unusual, but it runs a rather recent Linux kernel. It is capable of that only because I can choose which OS features to keep and which to discard; I would be unable to install a stock desktop Ubuntu on it. Under Vista that choice is not available, so no one can properly deliver benchmarks to demonstrate performance cost of ANY of its features, including the DRM. The closest one can manage would be to disable Aero and compare particular benchmarks against XP, but even then one cannot isolate the performance effects of individual Vista features.
To me it sounds like you are ultimately arguing that all new features in Vista are "free" if they are not "used". Since one cannot benchmark specific features in Vista, there is no way to prove you right or wrong.
What performance benefits do you think you're going to see by removing a feature that isn't being used ?
Since we'll never really know with Vista, I can only answer for Linux: I have created a kernel that has the bare minimum drivers (IDE, serial console, ext3, ptys) compiled in and boots directly to a BusyBox shell. On an embedded 486/100MHz computer, it can go from LILO's "Loading linux......" to a usable prompt in about 7 seconds. It loads only crond, inetd, sshd, and two user daemons. It's a bit minimal, but it accomplishes its job with only 32MB RAM and 100MB hard disk.
This argument is meaningless. It could be made about every single aspect of Vista that delivers some form of functionality you personally have no interest in.
How nice of you to notice that. Of course, some operating systems (not Vista) allow the user to REMOVE functions they have no interest in and gain the corresponding performance benefits.
You see, I am of the position that users haven't taken the equivalent step of "learning to drive" with respect to computers.
The equivalent of "learning to drive" is knowing which buttons lead to which responses and in general why. And to be blunt the software industry as a whole -- but Microsoft especially -- has made it incredibly difficult to do just that. Try to use Vista and Office 2007's ribbon after several years of 95/98/2k/XP and Office 97/XP/2003, or work on someone else's computer after they have changed the skins on half their apps, or just wait a few months for an important Internet site to change their interface. These are all similar to trying to drive a car with levers instead of a steering wheel and pedals.
What you're asking for is an intermediate position between "knowing how to drive a car" and "knowing how to design a car". Users are right to balk at that expectation.
I wish practically all software and web applications had a "novice mode" that was un-skinnable, conformed to some kind of CUA standard, and was guaranteed to remain the same with each new upgrade. The only system even claiming to shoot for that is OS X and it's not quite making it.
We had a RAID5 failure from a RAID Inc. (I'm naming them to shame them) array. It had 16 drives, 15 for the RAID and 1 hot spare. Turns out that this device beeps and turns on one global indicator light for any drive failure and pressing one button on the device to kill the beep also kills the indicator light. Second, the indicator and beep will be lost with a power cycle. Finally, the beep itself was not very loud and the indicator light was COVERED UP by a locking cover. In a university research group environment with high student turnover it is absurdly easy to miss the drive failure notices because there is only one chance to get the warning -- the drive bays do NOT have individual failure lights. If you own one of these in a very noisy server room, you had better make sure to do periodic checks and maybe set up email notification too because its indicators suck. Worst yet, when the RAID did finally go offline there was zero chance to rebuild it even after replicating the data on the failed drive because the controller would not use another physical drive in that virtual slot.
Moral to any RAID manufacturer out there: 1) ALWAYS put failure indicators on the individual drive bays. 2) When a user manages to duplicate the data on a failed drive to an identical backup via ddrescue, for god sakes let them put the old data on the new drive back into the array and bring it online.
However, I also don't think it's sustainable. Without some law requiring that these corporations keep wages at a set rate, the low rates will decrease the size of the workforce.
Population continues to grow and technology continues to make more jobs obsolete: both trends work to increase the number of people needing some kind of job to buy food. So why will low rates cause the size of the workforce to drop? Are you assuming that wages too low to support life will cause a massive die-off, which will shrink the labor pool?
Well, that's the basic misunderstanding. Most C programmers would interpret my snippet as follows:...
I interpret that code as "the memory at absolute location 3 will be assigned the value 4", which in most sane systems will segfault before the assert can be tested. Unless you had an & that got eaten, in which case you're interpretation isn't too far off.
Yep that's about the loss per car the big 3 have been having for the past couple years. I agree with you that the cost of the labor is what is sinking them.
No, you're saying that GM cannot make a car with zero-cost labor.
Why blame labor costs if GM cannot add enough value converting raw materials to a car that people will buy them?
My unit employs about 40 people making 50-100K on average. Let's be generous and say our labor cost is 40 X $200,000 = $8,000,000 per year. We still spend over $10 million per year on energy alone. But then again our (tangible physical) product earns about half a billion dollars per year.
For instance, to simply output a line to a command line in Java you're looking at System.out.println("output"); whereas with c++ (for instance) you have cout << "output" << endl; As someone who's teaching this stuff, the second is easier to explain in detail and doesn't rely on saying "don't worry what System.out is".
You're trading Java's (OO + packages) for C++'s (OO + operator overloading + namespaces). I don't see how that is any simpler.
Especially in the sciences, it disproportionately punishes people who find it hard to do complex error-free calculations against the clock, even if their understanding of the subject in question is just fine.
To be a slight devil's advocate, in my current job (chemical engineering) I find that having been able to survive the computation-intensive parts of the first two years in college enabled me to be able to model more complex systems with computer algebra systems (Maple, Matlab, etc.) than my less math-savvy counterparts. I'm finding myself establishing a reputation as the "model guy" who can make a practical optimized solution out of nebulous "that doesn't seem right" feelings. It also helps that my undergrad was CS so I can write custom code for the cases Maple isn't suited for.
OTOH, I agree completely that teaching subjects in depth for their own sakes should be left for majors in those subjects, such that science/engineering math should just be about three semesters covering only the practical aspects of trigonometry, calculus, linear algebra, diff eq, and PDE.
On the third hand, I also agree with the implicit argument you make that formal education really isn't necessary for 90% of the jobs out there. But I think that getting back to the point where a basic high school education is sufficient for a middle-class income would require more analysis from a Marxist perspective than the Western world is ready for.
They were bucking to get extra pay for one of their own or a whole new paid position created instead, and would rather see the work not get done at all than have it done by volunteers. (Same went for after-school tutoring.)
There is something to be said for this strategy though. If whatever it is volunteers want to do is really important, such that NOT doing it is really hurting the students, then the district really should be funding a position to handle it. If the district won't do that, then either they have broken priorities (which should be highlighted for the next school board election) or the work just really isn't that important.
Of course ANY strategy to change the schools away from race-to-the-bottom is likely to fail in reality since we haven't found a way for ANY organization to be both hierarchical and democratically-driven at the same time.
I don't understand why the colliding black holes would be a problem. The matter/anti-matter cores are inside the event horizon, so when the black holes merge their explosion would be inside the event horizon and the energy released by the explosion would be unable to go anywhere. Hence the "black hole".
Now if they were neutron stars it would be a different matter.
Slashdot Mirror
WOOSH!
Care to state your case for its falsity?
I've been to grad school. QED.
I'm an ex-IBMer too. Thanks for sharing this, it makes a lot more sense now.
Configuration files which change by user shouldn't be stored in Program Files, they should be in the user's profile directory.
I think the point is that configuration files that DON'T change by the user are still (by default) in the console user's profile directory.
Suppose I have something like IIS for the server, and I'm an admin user, and IIS is configured by a text file. I launch some kind of IIS configuration editor, which opens "C:\Program Files\IIS\foobaz.cfg" for edits. However, the file really being edited is "C:\Users\me\...\foobaz.cfg".
I think that's what the GP is talking about.
where some fool runs rm -rf as admin and it only stops deleting things when it deletes the delete command itself...
No it doesn't. Once 'rm' is executing it won't stop until all files it can delete are deleted.
You can also do "cat /dev/zero > /dev/hda" and watch the various daemons start dying and even some kernel oopses, yet 'cat' keep going.
Well, here's the problem, as I see it. The command line is an expressive interface, but largely limited to character-cell based interfaces. I just think we need to get out of that mode of thought. If you can take the command line out of simple character cells, that would make it even *more* expressive.
(Disclaimer: I wrote a console-based terminal emulator to bring BBS-era convenience to the Linux console. Works pretty well, and can be very convenient at times.)
I think the issue is that the character-cell metaphor is a "good enough" interface for mouse-less work, especially with ncurses encapsulating 30 years of text terminals development. You can still add mouse and go (T)IMP too: see the OSS port of Turbo Vision. This metaphor also works over the simplest available two-way links ala ssh/telnet/serial port/etc. Once you break out of the character-cell metaphor, you've got the full X11 GUI that works over network too. The question is why would you seek any kind of intermediate ground when both metaphors already work OK?
My personal opinion is that the next great interface revolution won't occur until either a) voice commands become the norm, or b) we can project a HUD-like screen over our real sight (total immersion / augmented reality). I suspect B will happen first, and it will probably come in a form like the Nintendo WII where multiple 3D pointers are used.
I would guess this is a Unicode issue. If your LANG is something like en_US.UTF-8 then switching it to en_US might fix things.
Strange. You're for both a flat tax and the Basic Income Guarantee. I've never seen that. Does your libertarian right hand want to cut off your socialist left hand?
Right. So the unemployment I took for six months in 2005 enslaved me such that I didn't finish my MS to get my shiny new physical engineering job.
Won't convince him. He'll cite Fannie and Freddie if he's a conservative wingnut, or the existence of the Fed if he's a libertard.
So what exactly IS "relevant" to this discussion? Only consumer desktop machines that might play HD video across HDMI from a Blu-ray source?
My embedded system is a bit unusual, but it runs a rather recent Linux kernel. It is capable of that only because I can choose which OS features to keep and which to discard; I would be unable to install a stock desktop Ubuntu on it. Under Vista that choice is not available, so no one can properly deliver benchmarks to demonstrate performance cost of ANY of its features, including the DRM. The closest one can manage would be to disable Aero and compare particular benchmarks against XP, but even then one cannot isolate the performance effects of individual Vista features.
To me it sounds like you are ultimately arguing that all new features in Vista are "free" if they are not "used". Since one cannot benchmark specific features in Vista, there is no way to prove you right or wrong.
What performance benefits do you think you're going to see by removing a feature that isn't being used ?
Since we'll never really know with Vista, I can only answer for Linux: I have created a kernel that has the bare minimum drivers (IDE, serial console, ext3, ptys) compiled in and boots directly to a BusyBox shell. On an embedded 486/100MHz computer, it can go from LILO's "Loading linux......" to a usable prompt in about 7 seconds. It loads only crond, inetd, sshd, and two user daemons. It's a bit minimal, but it accomplishes its job with only 32MB RAM and 100MB hard disk.
This argument is meaningless. It could be made about every single aspect of Vista that delivers some form of functionality you personally have no interest in.
How nice of you to notice that. Of course, some operating systems (not Vista) allow the user to REMOVE functions they have no interest in and gain the corresponding performance benefits.
You see, I am of the position that users haven't taken the equivalent step of "learning to drive" with respect to computers.
The equivalent of "learning to drive" is knowing which buttons lead to which responses and in general why. And to be blunt the software industry as a whole -- but Microsoft especially -- has made it incredibly difficult to do just that. Try to use Vista and Office 2007's ribbon after several years of 95/98/2k/XP and Office 97/XP/2003, or work on someone else's computer after they have changed the skins on half their apps, or just wait a few months for an important Internet site to change their interface. These are all similar to trying to drive a car with levers instead of a steering wheel and pedals.
What you're asking for is an intermediate position between "knowing how to drive a car" and "knowing how to design a car". Users are right to balk at that expectation.
I wish practically all software and web applications had a "novice mode" that was un-skinnable, conformed to some kind of CUA standard, and was guaranteed to remain the same with each new upgrade. The only system even claiming to shoot for that is OS X and it's not quite making it.
We had a RAID5 failure from a RAID Inc. (I'm naming them to shame them) array. It had 16 drives, 15 for the RAID and 1 hot spare. Turns out that this device beeps and turns on one global indicator light for any drive failure and pressing one button on the device to kill the beep also kills the indicator light. Second, the indicator and beep will be lost with a power cycle. Finally, the beep itself was not very loud and the indicator light was COVERED UP by a locking cover. In a university research group environment with high student turnover it is absurdly easy to miss the drive failure notices because there is only one chance to get the warning -- the drive bays do NOT have individual failure lights. If you own one of these in a very noisy server room, you had better make sure to do periodic checks and maybe set up email notification too because its indicators suck. Worst yet, when the RAID did finally go offline there was zero chance to rebuild it even after replicating the data on the failed drive because the controller would not use another physical drive in that virtual slot.
Moral to any RAID manufacturer out there: 1) ALWAYS put failure indicators on the individual drive bays. 2) When a user manages to duplicate the data on a failed drive to an identical backup via ddrescue, for god sakes let them put the old data on the new drive back into the array and bring it online.
I gotta wonder what the point of an ignorable notification is though. If I can ignore it, how does knowing about it help?
However, I also don't think it's sustainable. Without some law requiring that these corporations keep wages at a set rate, the low rates will decrease the size of the workforce.
Population continues to grow and technology continues to make more jobs obsolete: both trends work to increase the number of people needing some kind of job to buy food. So why will low rates cause the size of the workforce to drop? Are you assuming that wages too low to support life will cause a massive die-off, which will shrink the labor pool?
Well, that's the basic misunderstanding. Most C programmers would interpret my snippet as follows:...
I interpret that code as "the memory at absolute location 3 will be assigned the value 4", which in most sane systems will segfault before the assert can be tested. Unless you had an & that got eaten, in which case you're interpretation isn't too far off.
Yep that's about the loss per car the big 3 have been having for the past couple years. I agree with you that the cost of the labor is what is sinking them.
No, you're saying that GM cannot make a car with zero-cost labor.
Why blame labor costs if GM cannot add enough value converting raw materials to a car that people will buy them?
Welcome to the world of manufacturing.
My unit employs about 40 people making 50-100K on average. Let's be generous and say our labor cost is 40 X $200,000 = $8,000,000 per year. We still spend over $10 million per year on energy alone. But then again our (tangible physical) product earns about half a billion dollars per year.
They certainly are today when profit margins melted away and car prices are not competitive anymore due to way too high worker costs.
10% of the cost of a car is the labor. 10%. If the UAW magnanimously granted GM FREE LABOR that would bring your $30,000 truck price down to $27,000.
For instance, to simply output a line to a command line in Java you're looking at
System.out.println("output");
whereas with c++ (for instance) you have
cout << "output" << endl;
As someone who's teaching this stuff, the second is easier to explain in detail and doesn't rely on saying "don't worry what System.out is".
You're trading Java's (OO + packages) for C++'s (OO + operator overloading + namespaces). I don't see how that is any simpler.
Especially in the sciences, it disproportionately punishes people who find it hard to do complex error-free calculations against the clock, even if their understanding of the subject in question is just fine.
To be a slight devil's advocate, in my current job (chemical engineering) I find that having been able to survive the computation-intensive parts of the first two years in college enabled me to be able to model more complex systems with computer algebra systems (Maple, Matlab, etc.) than my less math-savvy counterparts. I'm finding myself establishing a reputation as the "model guy" who can make a practical optimized solution out of nebulous "that doesn't seem right" feelings. It also helps that my undergrad was CS so I can write custom code for the cases Maple isn't suited for.
OTOH, I agree completely that teaching subjects in depth for their own sakes should be left for majors in those subjects, such that science/engineering math should just be about three semesters covering only the practical aspects of trigonometry, calculus, linear algebra, diff eq, and PDE.
On the third hand, I also agree with the implicit argument you make that formal education really isn't necessary for 90% of the jobs out there. But I think that getting back to the point where a basic high school education is sufficient for a middle-class income would require more analysis from a Marxist perspective than the Western world is ready for.
They were bucking to get extra pay for one of their own or a whole new paid position created instead, and would rather see the work not get done at all than have it done by volunteers. (Same went for after-school tutoring.)
There is something to be said for this strategy though. If whatever it is volunteers want to do is really important, such that NOT doing it is really hurting the students, then the district really should be funding a position to handle it. If the district won't do that, then either they have broken priorities (which should be highlighted for the next school board election) or the work just really isn't that important.
Of course ANY strategy to change the schools away from race-to-the-bottom is likely to fail in reality since we haven't found a way for ANY organization to be both hierarchical and democratically-driven at the same time.
I don't understand why the colliding black holes would be a problem. The matter/anti-matter cores are inside the event horizon, so when the black holes merge their explosion would be inside the event horizon and the energy released by the explosion would be unable to go anywhere. Hence the "black hole".
Now if they were neutron stars it would be a different matter.