DOS software was what made me hate Microsoft
on
MS DOS: A Eulogy
·
· Score: 1
Specifically, Microsoft SmartDrive (their drive-cacheing software for DOS) and MS Flight Simulator 5.0. They aren't compatible. Tried to install FS5, bombed out during the install. I looked things over, and decided to deltree D:\FS5 and try again.
It took much longer than I thought it should have. I did a quick 'dir' and... yikes! Everything on D: was gone! I lost a bunch of stuff (fortunately nothing utterly irreplaceable). I reinstalled FS5, and sure enough, it bombed out again, and there was a crosslink from inside D:\FS5 to D:\.
I did some digging, and found out that if I disabled SmartDrive, it'd work fine. That's right, Microsoft Flight Simulator 5 is incompatible with... Microsoft SmartDrive.
The body works so well because it's highly highly highly redundant at the cellular level, not because there is a brilliant master control program controlling the most efficient implementation.
Well, yes and no. There is lots of redundancy in the human body (even perhaps too much for typical conditions nowadays in the case of the liver and kidneys) but there is also the autonomic nervous system, which is an active controller, and it's at that level that this is being proposed.
To get serious reliability you do need redundancy (RAID, failover, etc.) but you also need load balancing and monitoring software which works at a higher level. If your disk system fails you're screwed, but if you get hit with a Slashdot effect you can also be screwed unless you can dynamically shift resources around.
Of course, that kind of stuff is hard. Look at all the problems Linux itself has with out-of-memory conditions and all the different approaches that have been proposed for dealing with it. None is obviously the best way, and each has advantages and disadvantages. Of course, in all of them there's at least some overhead, so your point about capacity isn't completely irrelevant even at this level.
I was over at my parents' house on Oct. 6th and, of course, Sept. 11th came up. I tried to explain how stupid the notion of encryption with backdoors was, and how there had never been any verified case of messages hidden in images on the Internet.
Two of my aunts mentioned the coverage on ABC. They thought that the demonstration images shown had actually been found and related to the terrorist strikes. I didn't actually see the broadcast, but the two ladies involved aren't stupid. It must have been pretty misleading coverage to give them that impression.
Did anyone actually see the story when it was broadcast and can comment on it?
Re:Not to rain on anyone's parade...
on
Talking Palm
·
· Score: 1
A proper voice recognition system should be able to understand any words in the English language.
Can't be done with any computer today, and certainly not on a Palm. Human language depends too much on context and general background knowledge.
There's a (possibly apocryphal) story about how the speech recognition group at Microsoft is nicknamed the "wreck a nice beach" group. Say it out loud fast to understand why.
Picking out individual words in speech streams is hard. When you know a language it sounds to you like there are distinct gaps between words, but if you look on an oscilloscope, there aren't any. Think about times you've heard people speaking a language you didn't understand... it all sounded like a continuous stream, didn't it? You couldn't pick out any individual words, right?
Even humans need training and context to understand speech, and even humans get it wrong sometimes. As I said, no computer today has the necessary processing power and knowledge to do so.
On the other hand, in limited domains (e.g. words specified in advance, and/or voice specified in advance), quite a bit can be done. I can see how for the limited set of tasks a Palm is typically called on to perform, it might be effective. But I noted in the article that they had to add a co-processor. No 16-MHz Motorola 68000 has a chance in hell of running any useful speech recognition program.
Why do I get this mental picture of a relentless, unstoppable robotic alarm clock (with an Austrian accent) that's going to come find you and wake you up, no matter where you are?
It can't be snoozed. It can't be unplugged. It doesn't feel pity, or remorse, or fatigue. And it absolutely will not stop, ever, until you are up!
Seriously, I don't get it. What is supposed to be so cool about trolling? What makes you do it?
I mean, it just seems so stupid. What's the point? It's easy to act like a moron. Morons do it effortlessly. What does it prove to act stupid - that you can act dumb, too? This doesn't strike me as a grand accomplishment. Perhaps it's a malicious joy in infuriating others. Again, this doesn't seem particularly impressive - you can irritate people by accident. It's another no-brainer. What's the point?
Let me ask you something. Can you come up with something to say that would make readers feel as good as what you wrote made them angry?
Now, that would be impressive. Improving things is always harder than destroying them, whether moods or communication or software or buildings.
I know lots of people who know how to insult, and only one or two who knows how to give a really good backrub. Lots of people can break things, but only a few know how to repair them, much less make improvements. Do you believe that you can't post something that would be considered "Insightful", and so out of jealousy you want to ruin things for others?
So, again, I ask: What's the point? What do you get out of it?
I'd really like to see obfuscated/weird code in a language less "obfuscatable" than C (maybe Java or somethingorother?).
Just throwing obfuscated code constructs at the program isn't enough to win these days. (I know, I've tried.) To have a realistic chance of winning, your code has to excel in several ares of obfuscation (style, algorithm, abuse of the rules) and show economy and elegance and do something non-trivial.
Check out the winner last year that fit an entire X-Windows flight simulator into 1.5K of C code... can't link to it right now, but it's called "banks.c".
I'm going to enter again this year. I've got a program that I've progressively improved over several iterations. I just thought of a few ways to shave some bytes here and there; maybe I'll have enough room to squeeze in a graphical output routine...
The article says "SunnComm embeds a technology, called MediaCloq, into a CD to make the CD's directory structure invisible so it cannot be read by a personal computer." Later, however, there's the statement: "So they can make six copies; as long as their disc is in the tray of their computer, they can make those copies...
It appears that they also have some kind of downloadable music format; the player for it apparently checks to see if the CD is present.
Their website is maddeningly vague (understandably) and, to add insult to injury, it requires Flash. But there's a mention of using "Microsoft's DRM technology for music downloads".
A geosynchronous satellite stays more or less directly over some spot on the equator. I don't know if there's any international law/agreement on how high a country's jurisdiction extends. I suppose a country could at least claim that it exends up indefinitely.
Of course, by those rules the U.S. would own part of the Sun every day...
Can't happen. There's a reason it's called the Accelerated Graphics Port. It's a spec for a special port with priority access to memory and such. It was never designed as a "bus", just as a "port" to connect two points, and it's nearly impossible to put two into a system.
There's a reason why you've never seen a motherboard with two AGP slots. They can't exist, given the current spec.
What if I was to create a file consisting of nothing but the letter X that was, say, 1Gb in size, and leave it on my linux webserver with a name like "root.exe"?
I'm not a web guy, but don't most of the exploits that look for, e.g., CMD.EXE try to get the server (i.e. your box) to run it, not the client?
Now, you could create CGI's (or whatever) with the same names that the IIS exploits call (which are unlikely to already exist on an Apache system) and make them return, e.g., really large files or script code to crash the Windows box that's contacting yours.
And, yes, IANAL, but I can't see how anyone could have a leg to stand on to sue you or anything. They let their box become infected. Their box initiated the connection. It should be up to you how you want to respond.
If you're saying morals are absolute, you're saying that there's a universal law which mandates it, the only way that's possible is if there's a god doing the mandating.
Nope. Abbreviated version of a long argument: Consider chess. There is no rule of chess that says that you can't sacrifice your queen early in the game, but it's just about never a good idea if you want to win the game.
Analogy time: "rules of chess" -> "physical laws of the universe"; "strategic rules of chess (e.g. avoid sacrificing queen early)" -> "ethical/moral rules (e.g. avoid initiating violence".
If you want to be happy, you're going to need to cooperate with others, just as if you want to win at chess, you shouldn't throw away your queen.
Unless you have a flying carpet handy, I'd say the physical laws of the universe are absolute. I believe that ethical/moral laws follow as strategies given (a) human goals and (b) the constraints the universe places on us.
There is room for disagreement about exactly what strategies are best, but there are a few core rules that virtually all societies have had. (No random murder, no incest, etc.)
Perhaps we'll never know exactly what the optimal strategies are. But I think overall we're getting closer, or at least getting less wrong, no unlike science and its gradually-improving models of the physical world.
Besides, a God can't be the ultimate source of morality, anyway. Familiar with the Euthyphro Problem?
How is this freedom to prevent further attrocities? [...] The bottom line is that NEITHER approach works.
You have totally missed the point, though you came so close.
"Freedom" isn't supposed to "prevent further atrocities". It's simply valuable in its own right. It's not an "approach" that 'doesn't work' to prevent terrorism. It's almost completely orthogonal to the question of terrorism.
To make it plain...
Freedom is valuable in itself.
None of the proposed limits on our rights will make any significant difference in the chances for terrorism... but will greatly reduce our freedom.
Other approaches (e.g. a modified foreign policy, better intelligence, rational security measures) will significantly reduce the chances for terrorism without terrible consequences to our liberties.
Air pressure. A good chunk of the weight of the building is supported by internal air pressure. The buildings have to be nearly airtight anyway because of the difference between the atmospheric pressure between the ground and the top anyway, so they use air pressure to help support the weight.
Obviously there's a safety factor built in, so a couple of bronken windows wouldn't drop the towers to the ground. But with holes that big, and structural damage from the crashes themselves, they were going to collapse sooner or later.
The planes must have been flown by the terrorists into the buildings.
I was talking to my friend (a CFI, certified flight instructor) and he said that the commercial planes have autopilots that can be programmed for nearly any course desired. Getting 3 or more suicidal pilots is a lot harder than 3 or more suicidal people carrying instructions for how to program an autopilot.
Also, air traffic control should have been aware when the planes started heading for class B airspace, and scrambled interceptors. They've done it before. Of course, this wouldn't be the first time that plans weren't followed in a crisis.
I have to deal with VLB Every time I have to service an old, but usuable 486 machine.
Of course, VLB SCSI cards are few and far between. I've got a fully-decked-out VLB machine (just for the heck of it). It's got a VLB video card, and a VLB cacheing IDE controller, but though I've scrounged the dustbins for a long time, I've never even seen a VLB SCSI card.
Apparently they were made (Adaptec claims to have made some) but they seem to have all evaporated.
The Sherman Act doesn't regulate free trade, it regulates monopolies trying to use its monopoly power to expand into new markets. Period.
Actually, according to this primer on antitrust law, Section 2 of the Sherman Act makes it unlawful for a company to [...] maintain or acquire a monopoly position through unreasonable methods."
The OEM contract certainly sounds unreasonable to me, but, of course, IANAL.
If you want to actually play around with NPR, try NPRQuake, which is Quake as a cartoon. Really funky to play. Windows only, sadly, but gives you a very odd flavor. This was covered on Slashdot a few months ago.
The point-click-lock-you-in EULA has done away with the ability to have stable software on a computer...
The same type of reasoning used in EULAs is what allows Linux and other open-source software to disclaim liability for bugs and errors. If people were legally liable for the code they contribute to Linux, I rather suspect the whole project would have been stillborn.
Of course, you stated that "you should be held responsable[sic] for the quality if you are charging money for it". I can understand that, but how far would it go?
Computers are used in all kinds of medical equipment, but they are designed, hardware and software, from the ground up, as a whole. The manufacturers control (because they must) everything from the power plug on. And they go through endless tests. And because of all this, they are expensive and minimally customizable.
Applying similar standards to word processors would be folly; if your life depends on a word processor operating perfectly, you need to re-evaluate some of your decisions. Even looser standards would be a problem, though.
With increased liability, software quality would improve some. But open-source software, by its very nature, couldn't be warranted. And thus a lot of companies could get in trouble with shareholders and partners if they didn't use guaranteed software...
I agree that the current state of commercial software is deplorable, but a lot of careful thought is needed to figure out the best way to fix it.
There are some downsides to AIX:...
* Third-party support for AIX is sketchy. It is better to use IBM applications where possible.
I just had to write some modules to monitor some simple data about an AIX box. CPU, Memory Usage, Disk I/O, Page I/O, and Swap Usage for several platforms.
On Linux it was trivial. A few perl scripts to process some/proc files, and done. On Tru64 Unix, it wasn't much harder; just a call to table() and you've got the info you want.
On AIX it was a mess. You have to use nlist() to get the symbol table of the kernel, then open up the kernel image and lseek() and read() a raw data structure.
And that's for the documented items. You can't find the memory usage in any documented interface. You have to dig into an undocumented data structure (reverse engineered by the authors of "monitor") that changes between OS releases, often between point releases.
Sure, it's undocumented, and thus not guaranteed to change, so I can't really blame them for changing it. OK. But you can't, from a running C program, get the point release you're running under. You can call uname() and find out you're running under, say, 4.2, but you can't know if it's 4.2.0 or 4.2.1 that way... and it can matter.
You can fork out an run the shell command "oslevel", and then parse the output. All just to find out what version of AIX you're running on. All just to find out how the memory is being used! Sorry for the rant. AIX runs solid, and supports big systems. But programming it is way more exciting than it needs to be.
When I was an admin for a dorm network at the University of Michigan, we had one dorm router that would go down every weekend, without fail, on Saturday morning, and it wouldn't get fixed 'til Monday morning.
The tech would go in to the closet, move the brooms and buckets and ironing board and cleaning supplies out of the way, and find it had just spontaneously restarted and needed to be intitialized. It was like the power had failed, but no sign of any other problems, and if that circuit had failed it would have knocked out half the basement.
Eventually someone was in on Saturday morning, saw that it was down, and raced over to see what was up. I'm sure you can guess what he found...
Yup, this kid always did his laundry on Saturday morning, and he'd use the ironing board there to iron his clothes. And he'd unplug the router to plug in his iron, then plug it back in when he was done.
Competitions like the Perl obfuscated code contest are cool, but I'd like to see some contents based upon the best program than can be squeezed into an "antique" box like the TRS-80 Model I - that would really separate the hackers from the script kiddies...
Actually, most of those contests (e.g. the Obfuscated C Contest) have some very strict rules about how big the source can be, which does force economy of a sort. I takes some work to squeeze an X Window flight simulator into 1.5K of source code.
Just for the heck of it, I once wrote an ANSI C artificial life program. Doesn't sound too hard, but ANSI C only guarantees objects 32K in size! Anything more than that is not guaranteed to work. I also tageted it for low-memory environments like Minix where you can at most get 64K for code and 64K for data. The final program ran in less than 128K. It's really amazing to see how fast a program can run when it fits entirely into the L2 cache...
I used the first one and wrote the "joy-pci" driver in a few days. (Then Vojtech rewrote it to be half the size and more flexible.:->)
The first one really did cover (almost) everything you need to know. At the very least, after reading it newbies can ask much more intelligent questions. And having the kernel source to crib ideas from helps immensely, too.
I'd say this book, the kernel source, and a net connection to ask a few questions are all you need to write drivers for Linux. I'll get the second edition now, too. I've got a USB joystick and maybe a webcam that I'd like to add support for...
Slashdot Mirror
It took much longer than I thought it should have. I did a quick 'dir' and... yikes! Everything on D: was gone! I lost a bunch of stuff (fortunately nothing utterly irreplaceable). I reinstalled FS5, and sure enough, it bombed out again, and there was a crosslink from inside D:\FS5 to D:\.
I did some digging, and found out that if I disabled SmartDrive, it'd work fine. That's right, Microsoft Flight Simulator 5 is incompatible with... Microsoft SmartDrive.
At that point I started using Linux in earnest.
Well, yes and no. There is lots of redundancy in the human body (even perhaps too much for typical conditions nowadays in the case of the liver and kidneys) but there is also the autonomic nervous system, which is an active controller, and it's at that level that this is being proposed.
To get serious reliability you do need redundancy (RAID, failover, etc.) but you also need load balancing and monitoring software which works at a higher level. If your disk system fails you're screwed, but if you get hit with a Slashdot effect you can also be screwed unless you can dynamically shift resources around.
Of course, that kind of stuff is hard. Look at all the problems Linux itself has with out-of-memory conditions and all the different approaches that have been proposed for dealing with it. None is obviously the best way, and each has advantages and disadvantages. Of course, in all of them there's at least some overhead, so your point about capacity isn't completely irrelevant even at this level.
Two of my aunts mentioned the coverage on ABC. They thought that the demonstration images shown had actually been found and related to the terrorist strikes. I didn't actually see the broadcast, but the two ladies involved aren't stupid. It must have been pretty misleading coverage to give them that impression.
Did anyone actually see the story when it was broadcast and can comment on it?
Can't be done with any computer today, and certainly not on a Palm. Human language depends too much on context and general background knowledge. There's a (possibly apocryphal) story about how the speech recognition group at Microsoft is nicknamed the "wreck a nice beach" group. Say it out loud fast to understand why.
Picking out individual words in speech streams is hard. When you know a language it sounds to you like there are distinct gaps between words, but if you look on an oscilloscope, there aren't any. Think about times you've heard people speaking a language you didn't understand... it all sounded like a continuous stream, didn't it? You couldn't pick out any individual words, right? Even humans need training and context to understand speech, and even humans get it wrong sometimes. As I said, no computer today has the necessary processing power and knowledge to do so.
On the other hand, in limited domains (e.g. words specified in advance, and/or voice specified in advance), quite a bit can be done. I can see how for the limited set of tasks a Palm is typically called on to perform, it might be effective. But I noted in the article that they had to add a co-processor. No 16-MHz Motorola 68000 has a chance in hell of running any useful speech recognition program.
It can't be snoozed. It can't be unplugged. It doesn't feel pity, or remorse, or fatigue. And it absolutely will not stop, ever, until you are up!
I mean, it just seems so stupid. What's the point? It's easy to act like a moron. Morons do it effortlessly. What does it prove to act stupid - that you can act dumb, too? This doesn't strike me as a grand accomplishment. Perhaps it's a malicious joy in infuriating others. Again, this doesn't seem particularly impressive - you can irritate people by accident. It's another no-brainer. What's the point?
Let me ask you something. Can you come up with something to say that would make readers feel as good as what you wrote made them angry? Now, that would be impressive. Improving things is always harder than destroying them, whether moods or communication or software or buildings.
I know lots of people who know how to insult, and only one or two who knows how to give a really good backrub. Lots of people can break things, but only a few know how to repair them, much less make improvements. Do you believe that you can't post something that would be considered "Insightful", and so out of jealousy you want to ruin things for others?
So, again, I ask: What's the point? What do you get out of it?
Just throwing obfuscated code constructs at the program isn't enough to win these days. (I know, I've tried.) To have a realistic chance of winning, your code has to excel in several ares of obfuscation (style, algorithm, abuse of the rules) and show economy and elegance and do something non-trivial.
Check out the winner last year that fit an entire X-Windows flight simulator into 1.5K of C code... can't link to it right now, but it's called "banks.c".
I'm going to enter again this year. I've got a program that I've progressively improved over several iterations. I just thought of a few ways to shave some bytes here and there; maybe I'll have enough room to squeeze in a graphical output routine...
So, basically, the law hasn't really been set yet. Add in the fact that there's no generally accepted definition of 'outer space' and the situtation's pretty messed up.
It appears that they also have some kind of downloadable music format; the player for it apparently checks to see if the CD is present. Their website is maddeningly vague (understandably) and, to add insult to injury, it requires Flash. But there's a mention of using "Microsoft's DRM technology for music downloads".
Of course, by those rules the U.S. would own part of the Sun every day...
Can't happen. There's a reason it's called the Accelerated Graphics Port. It's a spec for a special port with priority access to memory and such. It was never designed as a "bus", just as a "port" to connect two points, and it's nearly impossible to put two into a system.
There's a reason why you've never seen a motherboard with two AGP slots. They can't exist, given the current spec.
I'm not a web guy, but don't most of the exploits that look for, e.g., CMD.EXE try to get the server (i.e. your box) to run it, not the client?
Now, you could create CGI's (or whatever) with the same names that the IIS exploits call (which are unlikely to already exist on an Apache system) and make them return, e.g., really large files or script code to crash the Windows box that's contacting yours.
And, yes, IANAL, but I can't see how anyone could have a leg to stand on to sue you or anything. They let their box become infected. Their box initiated the connection. It should be up to you how you want to respond.
Nope. Abbreviated version of a long argument: Consider chess. There is no rule of chess that says that you can't sacrifice your queen early in the game, but it's just about never a good idea if you want to win the game. Analogy time: "rules of chess" -> "physical laws of the universe"; "strategic rules of chess (e.g. avoid sacrificing queen early)" -> "ethical/moral rules (e.g. avoid initiating violence".
If you want to be happy, you're going to need to cooperate with others, just as if you want to win at chess, you shouldn't throw away your queen.
Unless you have a flying carpet handy, I'd say the physical laws of the universe are absolute. I believe that ethical/moral laws follow as strategies given (a) human goals and (b) the constraints the universe places on us.
There is room for disagreement about exactly what strategies are best, but there are a few core rules that virtually all societies have had. (No random murder, no incest, etc.) Perhaps we'll never know exactly what the optimal strategies are. But I think overall we're getting closer, or at least getting less wrong, no unlike science and its gradually-improving models of the physical world.
Besides, a God can't be the ultimate source of morality, anyway. Familiar with the Euthyphro Problem?
"Climate is what we expect. Weather is what we get."
You have totally missed the point, though you came so close.
"Freedom" isn't supposed to "prevent further atrocities". It's simply valuable in its own right. It's not an "approach" that 'doesn't work' to prevent terrorism. It's almost completely orthogonal to the question of terrorism.
To make it plain...
- Freedom is valuable in itself.
- None of the proposed limits on our rights will make any significant difference in the chances for terrorism... but will greatly reduce our freedom.
- Other approaches (e.g. a modified foreign policy, better intelligence, rational security measures) will significantly reduce the chances for terrorism without terrible consequences to our liberties.
Ergo... don't allow warrantless wiretaps et. al.!Air pressure. A good chunk of the weight of the building is supported by internal air pressure. The buildings have to be nearly airtight anyway because of the difference between the atmospheric pressure between the ground and the top anyway, so they use air pressure to help support the weight.
Obviously there's a safety factor built in, so a couple of bronken windows wouldn't drop the towers to the ground. But with holes that big, and structural damage from the crashes themselves, they were going to collapse sooner or later.
I was talking to my friend (a CFI, certified flight instructor) and he said that the commercial planes have autopilots that can be programmed for nearly any course desired. Getting 3 or more suicidal pilots is a lot harder than 3 or more suicidal people carrying instructions for how to program an autopilot.
Also, air traffic control should have been aware when the planes started heading for class B airspace, and scrambled interceptors. They've done it before. Of course, this wouldn't be the first time that plans weren't followed in a crisis.
Of course, VLB SCSI cards are few and far between. I've got a fully-decked-out VLB machine (just for the heck of it). It's got a VLB video card, and a VLB cacheing IDE controller, but though I've scrounged the dustbins for a long time, I've never even seen a VLB SCSI card.
Apparently they were made (Adaptec claims to have made some) but they seem to have all evaporated.
Actually, according to this primer on antitrust law, Section 2 of the Sherman Act makes it unlawful for a company to [...] maintain or acquire a monopoly position through unreasonable methods."
The OEM contract certainly sounds unreasonable to me, but, of course, IANAL.
If you want to actually play around with NPR, try NPRQuake, which is Quake as a cartoon. Really funky to play. Windows only, sadly, but gives you a very odd flavor. This was covered on Slashdot a few months ago.
The same type of reasoning used in EULAs is what allows Linux and other open-source software to disclaim liability for bugs and errors. If people were legally liable for the code they contribute to Linux, I rather suspect the whole project would have been stillborn.
Of course, you stated that "you should be held responsable[sic] for the quality if you are charging money for it". I can understand that, but how far would it go?
Computers are used in all kinds of medical equipment, but they are designed, hardware and software, from the ground up, as a whole. The manufacturers control (because they must) everything from the power plug on. And they go through endless tests. And because of all this, they are expensive and minimally customizable.
Applying similar standards to word processors would be folly; if your life depends on a word processor operating perfectly, you need to re-evaluate some of your decisions. Even looser standards would be a problem, though.
With increased liability, software quality would improve some. But open-source software, by its very nature, couldn't be warranted. And thus a lot of companies could get in trouble with shareholders and partners if they didn't use guaranteed software...
I agree that the current state of commercial software is deplorable, but a lot of careful thought is needed to figure out the best way to fix it.
I just had to write some modules to monitor some simple data about an AIX box. CPU, Memory Usage, Disk I/O, Page I/O, and Swap Usage for several platforms.
On Linux it was trivial. A few perl scripts to process some /proc files, and done. On Tru64 Unix, it wasn't much harder; just a call to table() and you've got the info you want.
On AIX it was a mess. You have to use nlist() to get the symbol table of the kernel, then open up the kernel image and lseek() and read() a raw data structure.
And that's for the documented items. You can't find the memory usage in any documented interface. You have to dig into an undocumented data structure (reverse engineered by the authors of "monitor") that changes between OS releases, often between point releases.
Sure, it's undocumented, and thus not guaranteed to change, so I can't really blame them for changing it. OK. But you can't, from a running C program, get the point release you're running under. You can call uname() and find out you're running under, say, 4.2, but you can't know if it's 4.2.0 or 4.2.1 that way... and it can matter.
You can fork out an run the shell command "oslevel", and then parse the output. All just to find out what version of AIX you're running on. All just to find out how the memory is being used! Sorry for the rant. AIX runs solid, and supports big systems. But programming it is way more exciting than it needs to be.
The tech would go in to the closet, move the brooms and buckets and ironing board and cleaning supplies out of the way, and find it had just spontaneously restarted and needed to be intitialized. It was like the power had failed, but no sign of any other problems, and if that circuit had failed it would have knocked out half the basement.
Eventually someone was in on Saturday morning, saw that it was down, and raced over to see what was up. I'm sure you can guess what he found...
Yup, this kid always did his laundry on Saturday morning, and he'd use the ironing board there to iron his clothes. And he'd unplug the router to plug in his iron, then plug it back in when he was done.
Actually, most of those contests (e.g. the Obfuscated C Contest) have some very strict rules about how big the source can be, which does force economy of a sort. I takes some work to squeeze an X Window flight simulator into 1.5K of source code.
Just for the heck of it, I once wrote an ANSI C artificial life program. Doesn't sound too hard, but ANSI C only guarantees objects 32K in size! Anything more than that is not guaranteed to work. I also tageted it for low-memory environments like Minix where you can at most get 64K for code and 64K for data. The final program ran in less than 128K. It's really amazing to see how fast a program can run when it fits entirely into the L2 cache...
The first one really did cover (almost) everything you need to know. At the very least, after reading it newbies can ask much more intelligent questions. And having the kernel source to crib ideas from helps immensely, too.
I'd say this book, the kernel source, and a net connection to ask a few questions are all you need to write drivers for Linux. I'll get the second edition now, too. I've got a USB joystick and maybe a webcam that I'd like to add support for...