yes, I understand the chemistry behind it, and with the current configuration of the cell, and the compounds they use, they are stuck at 1.2 volts. However, if they change the compound, or change the cell arrangement, they could change the voltage to any reasonable number fairly easily. So the question is, why haven't they?
The problem with most rechargables is that the voltage is less than with alkalines. A single AAA, AA, C, or D battery should be 1.5 volts, but all NiCd's and most NiMH I've seen are 1.2 volts, which means that whatever device you are using has to draw more current (amps) from the battery, and the battery either won't last as long (for low power), or you may pull too much current and have problems (for really high power devices). I don't understand why rechargeables have a lower voltage, and at this point it's probably just because they've always been that way.
Anyways, if you need a good quality battery that lasts a long time, I'd suggest either modifying the device you have to use a known Li+ (Lithium-Ion) battery, or go to your local custom battery shop and have some made in the form factor you need. For low power devices, you can probably get longer life from a charge on a Li+ battery than alkalines, for the same or less weight. My sony minidisc walkman uses a fairly old one, about half the size and weight of a single AA, and it powers the minidisc player for over 10 hours per charge.
Of course, when shopping, look at the mAH (milli-Amp*Hours, a AA is around 2500-2800 for low power, much lower for high-power) of the batteries, but take into account the type of device you are using. High-power devices usually favor Li+ or NiMH over NiCd or Alkaline. This is a good site for battery myths that should help, but they don't cover the (superior) Li+ batteries. If you really need the power, use Li+, nothing beats them for power/size/weight.
exactly. I've been waiting for this sort of thing, but I'm waiting on the new Samsung I500 to come down in price. I get calls all the time that require me to log in to a system, but for now I just use my ThinkPad and the data cable for my phone. (Funny thing is, if you have Sprint PCS Vision and a data cable, you get unlimited internet for no extra charge... The only problem is, as soon as they went to the unlimited plans, they stopped selling the cables, so you have to find them on ebay)
Have you considered using two printers? It obviously depends on your printing needs, but I know that I rarely have the need to print color documents (only for documents that are used in a meeting or design specification). I use an old HP LaserJet 4 (old, but these things are built like tanks and toner is extremely cost effective) that I bought at a government auction for $7.50 (including toner cartridge, 3 monitors, and a UPS) for the majority of my printing. It's fast, reliable, and so far has printed over 3000 pages for me, still on the toner cartridge that was in it when I picked it up. For color printing, I use a Canon S600, which has separate (but kinda small) ink tanks. Of course, ink for the Canon is not cheap, but the cartridges are easily refillable, the print quality is excellent, and it's as fast as you should really go with an inkjet, since it can sometimes smear the previous page when it kicks one out if you use cheap paper. If you can manage to pick a good LJ up at auction (any or the larger 4, 5, or 6 series is good for volume work, 6L is decent for desktop), you could buy both printers and plenty of toner and ink for under the $500 you mentioned. When you need color, use the inkjet (Parallel or USB), when you need economy and speed, use the laser (Parallel, Serial, Network, some models USB).
Re:Happy Birthday to my favorite distro
on
Slackware Turns 10
·
· Score: 2, Interesting
Yeah, I think I may have even used this distro at some point. I had used Coherent before (a cheap intel *nix), and wanted to try out the "new Linux thing". God, I was a geek for a 13-year old. Anyways, last version I used was slack 95, then went to RH, and now gentoo. I guess I've just moved through the best package management schemes, since slack was a big deal with all the packages available for it (tgz), then RedHat's rpm, but gentoo's portage tops them all. There's nothing quite like typing "emerge sync;emerge -u --deep world" and recompiling every updated package for your system, with all the optimizations you want. In my opinion, slackware was the one that really got a lot of people to try Linux (sorry, GNU/Linux, whatever...) and is probably responsible for the creation of a lot of the distros we have today. As needs have changed for different groups, they split off, but I think Slackware got them to look at Linux in the first place. People have different needs/abilities, and now there are distros to match most needs pretty well. Happy Birthday Slack!
As I'll be graduating from UCF this fall with a CS degree, I suppose I'm qualified to answer... On with it then... First of all, if you're really having trouble with the class, it's probably best to seek help that's actually breathing, as a book often fails to give that last bit of insight that's keeping you from understanding the methods. Second, don't worry about the foundation exam, if you know the basics, and can do common proofs from discrete, it's actually quite easy. As for the book, I took Intro to Discrete in 1999, and the book we used then was excellent, James Hein's "Discrete Structures, Logic, and Computability". If automata and languages are the ones giving you trouble after reading that book, check out the upper level (COT4210) book, Sudkamp's "Languages & Machines". The first book should give you plenty for that class, though... Oh, and one more free tip: when you take 4210, don't take Torosolu's class, try for (Drs.) Llewellyn, Dutton, Workman, or Guha (Arup)'s class. Actually, in all cases, try to get those professors...
The important factor here is how efficiently the signal can pass through a medium, and how directional the signal is. We don't live in a vacuum (if you didn't know this, you might need to get out once in a while:P), and therefore any signal sent has to travel through any objects in its path, including the air. Not sure exactly off hand which law of physics covers this, but lower frequencies lose less energy when passing through a medium, and are also less directional. Think about sound here, a low frequency (bass), say 40Hz will have relatively no problem passing through an object, like a car driving in front of your house at 2:00A.M., but a high frequency (treble), for example 10kHz is highly directional and has great difficulty traveling through objects. I believe this is due to the fact that to pass a signal through a medium, the particles in the medium must be excited to the same frequency as the signal, and it takes significantly less energy to excite a material at a lower frequency than it does at a higher one. Higher frequencies give higher fidelity, but are less resistant to signal loss.
I generally use Visio for this sort of thing, but planning out a large network using an automated software tool is impossible, at least in my experience. If you can logically group your network at some level, you will find that it will make planning, and/or diagramming your network much easier. Even the largest networks are broken down into several groups at some level, and you have to keep those groups in mind. You should know the usage patterns of these groups, how much bandwidth each group needs, and where they connect to, etc. Logically map out the best performance scenario for these groups. Then start with your smallest groups, and find similarities to form larger groups, until you have encompassed the entire network. I'm assuming that there are probably departmental connections needed, as well as outside connections, but the key is how much each group of people use each resource available. Many "well-planned" networks fail because they aimed to provide all resources evenly, but in the real-world, this is not the case. Most of the time, a certain group of people use specific resources, and use very little of other resources. Also keep in mind that you must be able to adapt if the resource requirements for a group change, because it will happen. It's the same thing as trying to use a database wizard to optimize your database. A computer just can't do that for you. The software doesn't have all the necessary input parameters to the problem, and even if it did, the problem is NP Hard. So, in reality, it doesn't matter what tool you use to diagram the network, but actual design still has to be done by human insight, or at least a well-trained monkey...
Ok first of all, regardless of who you ship with, get insurance, and make sure that the person receiving the package has to sign for it, preferably while providing ID. Airborne in my experience has been the best here, as the UPS driver just leaves any package in front of the door, rings the doorbell, and runs back to his truck. If you run to the door, you may catch him leaving, but if you aren't home, anyone could pick up your package. Either way, you should make sure you are there to pick up the package when it arrives, check the box requiring signature, insure it, and if convenient, tell the shipper to hold your item for pickup at their location. If they lose it at their location, you win. If you didn't sign for it, you win. All shippers have some bad drivers/handlers, and even if you know the particular driver for the destination, you should ALWAYS protect yourself from theft, damage, or other problems. Everything will probably be fine, but when it isn't, those who were the most cautious get their stuff back, and those who said "ship this here" will have a hard time getting anything back.
Another interesting point. I was reading a story on Yahoo about converting to Linux from Windows. Ernie Ball (a company that primarily makes music hardware/parts/guitar strings) was ordered to pay $90,000 to Microsoft within 120 days to be compliant (not enough licenses). The company used the court ordered 120 day compliance period to evaluate and install Linux, and from the article, never paid a dime.
The same thing could easily happen in this case. IF there is SCO IP in Linux, whatever code it is can be changed in minutes or hours, and even IF SCO was right, they might not get anything! Of course the other good example is USL vs. BSDI, but SCO apparently never paid attention in their history or business law classes...
Evidence? Where? So far, SCO and (maybe) IBM notwithstanding, I believe only a few people have seen ANY evidence in this case.
I think what you meant is "this seems like just as much evidence as SCO has presented", which is true. This equation should help. Let S be all evidence SCO has presented for public viewing. Let P be all public evidence presented against SCO. S = P = lim x->infinity (1/x).
No, a computer language is a tool. Choice of language is a combination of finding which tool is best suited to the task at hand, and personal preference. It's the same choice you'd encounter trying to drive a nail into a board. Writing directly in machine language is like driving the nail in with your forehead, ASM is like using the handle of a screwdriver, and pretty much every other language is like using a hammer. It really doesn't matter what brand of hammer you use, they all will do the task for you, although with varying degrees of effort.
The question is, if you chose hammer x over hammer y to drive in your nail, but you miss the nail and split open your thumb, is it the hammer's fault? Bad coding is like hitting yourself with a hammer, you've got nobody to blame but yourself, but everyone always yells "stupid f&%$ing hammer!"
Not exactly. Assuming that the hardware is ok, you can prove that a system is reliable for any given finite input (including, most importantly, all possible finite substrings of inputs, however it is not possible to test all possible inputs, since a portion of those are infinite), it's just that doing so in large systems takes enormous amounts of time, and of course, time = money. Take Microsoft, for example. It takes a team years to develop a product like Windows XP, run a few test cases, and fix the major bugs. But just think how long it would take to go through every possible input substring of a given length (and by substring/string I am including non-character inputs [mouse, network, etc]).
Consider a simple program that inputs 10 short strings of text and does some computations on those strings. Say for example that the system that has only a keyboard as input, that all input functions are guaranteed only to input A-Z (caps only), the space bar, and 0-9 (regex ((A-Z)*(0-9)*)*( )*), not to overflow, and that there are 10 inputs with exactly 10 characters for each input (spaces fill end of string). This means that there are 37 possibilities for each digit, totaling 37^100 unique possible inputs, about 6.61E156 possibilities, each 100 characters. Typing a million characters per second would take 2.094E145 years! Keep in mind that this is an extremely simple system.
Therefore, it is not possible to test ALL input cases of any nontrivial program, only a few selected cases, which most will agree is far from proving a program correct. Instead, developers should have detailed mathematical descriptions of how a program is to behave at each incremental step, and verify that the program follows those descriptions accurately. Programs can only be proven correct in the same manner that any discrete mathematic concept can be proven correct, with one of the most common methods of a functionality proof being mathematical induction. Based on a few basic assumptions (like that the functions you call work as documented), the rest of the system can be proven by proving the trivial parts and cases first, and then constructing a complete proof based on the trivial parts.
The problem with this is that a small change can have a big impact on the proof, and nobody actually takes the time to verify that everything still works. Companies don't often spend money on making their software 100% correct, they just need to add the nifty new features that their customers want before their competitors do. I'd be willing to bet that 90% of the bugs found in XP can be traced to a "nifty new feature" that broke code that may have been proven correct at some point.
In other words, the short answer is yes, if you can test every state, you can prove a program correct, but since that's usually impossible, it becomes the developers' responsibility to incrementally prove the system, which is far easier if all functionality is planned ahead of time, but still too time/money consuming for most software companies to bother with. Microsoft doesn't care if your computer crashes, you'll probably still pay them, and as much as I'd like to think otherwise, OSS isn't much different (although it's usually more time than money there).
It is nice to have a big standard library, but you also have to consider what is included and what isn't. What often happens (MIL STD 1812 - The Ada Language) is that you get a HUGE library of functions, yet someone can still come up with a few routines that 50% of your programs will use, but aren't included in the library. Also, you may only use 3 or 4 of the library functions, out of the thousands available in such a language. This wastes space. Also, at some point you may need to control not only what your code does, but how it does it. Ever written something for an embedded system or special hardware in Java, Perl, or Python? Didn't think so. C is as popular as it is for two reasons: 1) Nice syntactic properties, simple language (which is probably why the majority of other languages use a syntax like C, which of course is actually derived from ALGOL). and 2) It allows very low level access to the machine, thus letting programmers have something to read which can easily be translated to assembly or object code (I believe all keywords and operators map to a single opcode on most architectures, excluding load/store operations and accumulator machines).
In other words, if you don't like C/C++, don't do anything that requires control over what happens. Hell, while you're at it, why not just use Visual Basic to write all your apps, after all, we all know the Microsoft libraries are (HEAVY)(SARCASM)flawless.(/SARCASM)(/HEAVY)
I agree, though it doesn't necessarily need to cover the differences, just maybe tout a few of the traits of the BSD's and OpenBSD specifically. One easy example is the secuity issue. I also think some history from at least 4.4BSD should be covered, since "The Design and Implementation of the 4.4BSD Operating System" is an excellent book and covers the earlier history. Definitely should point out that the OpenBSD camp leans heavily towards security and stability in general.
The book should also cover topics on the basics, configuring services to be run at init, but should avoid trying to be a "Complete Reference", since no book ever really is. Just some basic administrative tasks, things that any sysadmin would do on a regular basis (configuring, building, and testing a new kernel), and the things none of us ever really want to have to do (Recovery of bad slices). Hardware configuration should also be covered, as that is certainly one of the more difficult things for the Win95 and up generation. (What, you mean I actually have to know what I have inside that shiny metal box thingy?) Leave the specifics of major services (DNS, Web, *mail, etc...) up to the books written for them, just overview the services, and give sysadmins a good reference.
That reminds me... If anyone has ever read the Coherent (circa 1994) manuals, those were excellent, and I wish there were something like that now. Complete command references, system architecture overviews, good stuff that is useful to admins and newbies.
The software errors/stupidity that let the alleged criminals find out they were being tapped, or the fact that we don't have those errors here in the US. Not that I do anything wrong, but I'd still prefer not to be monitored by the government.
This reminds me of a/. article a while back about the US taking bids for a central repository of personal information. It may sound like a great idea to the politicians, but after a while they're going to need money, and guess what?... Then everyone's personal information is up for sale. Telcos have been doing this for a while, even my university (UCF) does this, and I get a few dozen porn/marketing spams a day, just because they have my email address in their records.
And you thought spam was a problem now....
Yeah, this is great. Now we'll have MORE stories from Microsoft employees about how great their lives are now that they've switched to Frontpage and Passport. Who knows, maybe this time they won't use a stock photo, too!
Microsoft: Un*x administrator switches to Passport, says 'much more reliable'
AP (next day): Microsoft 'Un*x administrator' found, works for Microsoft PR firm...
sound familiar?
Slashdot Mirror
yes, I understand the chemistry behind it, and with the current configuration of the cell, and the compounds they use, they are stuck at 1.2 volts. However, if they change the compound, or change the cell arrangement, they could change the voltage to any reasonable number fairly easily. So the question is, why haven't they?
>1TB capacities commonplace tomorrow!? Wow!!
Damn right, wow. That's a lot of MP3's. What's the RIAA going to do then?
The problem with most rechargables is that the voltage is less than with alkalines. A single AAA, AA, C, or D battery should be 1.5 volts, but all NiCd's and most NiMH I've seen are 1.2 volts, which means that whatever device you are using has to draw more current (amps) from the battery, and the battery either won't last as long (for low power), or you may pull too much current and have problems (for really high power devices). I don't understand why rechargeables have a lower voltage, and at this point it's probably just because they've always been that way.
Anyways, if you need a good quality battery that lasts a long time, I'd suggest either modifying the device you have to use a known Li+ (Lithium-Ion) battery, or go to your local custom battery shop and have some made in the form factor you need. For low power devices, you can probably get longer life from a charge on a Li+ battery than alkalines, for the same or less weight. My sony minidisc walkman uses a fairly old one, about half the size and weight of a single AA, and it powers the minidisc player for over 10 hours per charge.
Of course, when shopping, look at the mAH (milli-Amp*Hours, a AA is around 2500-2800 for low power, much lower for high-power) of the batteries, but take into account the type of device you are using. High-power devices usually favor Li+ or NiMH over NiCd or Alkaline. This is a good site for battery myths that should help, but they don't cover the (superior) Li+ batteries. If you really need the power, use Li+, nothing beats them for power/size/weight.
exactly. I've been waiting for this sort of thing, but I'm waiting on the new Samsung I500 to come down in price. I get calls all the time that require me to log in to a system, but for now I just use my ThinkPad and the data cable for my phone. (Funny thing is, if you have Sprint PCS Vision and a data cable, you get unlimited internet for no extra charge... The only problem is, as soon as they went to the unlimited plans, they stopped selling the cables, so you have to find them on ebay)
damnit, you beat me to it. I just saw the article, and that was the first thing that popped into my mind.
Have you considered using two printers? It obviously depends on your printing needs, but I know that I rarely have the need to print color documents (only for documents that are used in a meeting or design specification).
I use an old HP LaserJet 4 (old, but these things are built like tanks and toner is extremely cost effective) that I bought at a government auction for $7.50 (including toner cartridge, 3 monitors, and a UPS) for the majority of my printing. It's fast, reliable, and so far has printed over 3000 pages for me, still on the toner cartridge that was in it when I picked it up.
For color printing, I use a Canon S600, which has separate (but kinda small) ink tanks. Of course, ink for the Canon is not cheap, but the cartridges are easily refillable, the print quality is excellent, and it's as fast as you should really go with an inkjet, since it can sometimes smear the previous page when it kicks one out if you use cheap paper.
If you can manage to pick a good LJ up at auction (any or the larger 4, 5, or 6 series is good for volume work, 6L is decent for desktop), you could buy both printers and plenty of toner and ink for under the $500 you mentioned. When you need color, use the inkjet (Parallel or USB), when you need economy and speed, use the laser (Parallel, Serial, Network, some models USB).
Yeah, I think I may have even used this distro at some point. I had used Coherent before (a cheap intel *nix), and wanted to try out the "new Linux thing". God, I was a geek for a 13-year old. Anyways, last version I used was slack 95, then went to RH, and now gentoo. I guess I've just moved through the best package management schemes, since slack was a big deal with all the packages available for it (tgz), then RedHat's rpm, but gentoo's portage tops them all. There's nothing quite like typing "emerge sync;emerge -u --deep world" and recompiling every updated package for your system, with all the optimizations you want. In my opinion, slackware was the one that really got a lot of people to try Linux (sorry, GNU/Linux, whatever...) and is probably responsible for the creation of a lot of the distros we have today. As needs have changed for different groups, they split off, but I think Slackware got them to look at Linux in the first place. People have different needs/abilities, and now there are distros to match most needs pretty well.
Happy Birthday Slack!
As I'll be graduating from UCF this fall with a CS degree, I suppose I'm qualified to answer... On with it then... First of all, if you're really having trouble with the class, it's probably best to seek help that's actually breathing, as a book often fails to give that last bit of insight that's keeping you from understanding the methods. Second, don't worry about the foundation exam, if you know the basics, and can do common proofs from discrete, it's actually quite easy. As for the book, I took Intro to Discrete in 1999, and the book we used then was excellent, James Hein's "Discrete Structures, Logic, and Computability". If automata and languages are the ones giving you trouble after reading that book, check out the upper level (COT4210) book, Sudkamp's "Languages & Machines". The first book should give you plenty for that class, though... Oh, and one more free tip: when you take 4210, don't take Torosolu's class, try for (Drs.) Llewellyn, Dutton, Workman, or Guha (Arup)'s class. Actually, in all cases, try to get those professors...
The important factor here is how efficiently the signal can pass through a medium, and how directional the signal is. We don't live in a vacuum (if you didn't know this, you might need to get out once in a while :P), and therefore any signal sent has to travel through any objects in its path, including the air. Not sure exactly off hand which law of physics covers this, but lower frequencies lose less energy when passing through a medium, and are also less directional. Think about sound here, a low frequency (bass), say 40Hz will have relatively no problem passing through an object, like a car driving in front of your house at 2:00A.M., but a high frequency (treble), for example 10kHz is highly directional and has great difficulty traveling through objects.
I believe this is due to the fact that to pass a signal through a medium, the particles in the medium must be excited to the same frequency as the signal, and it takes significantly less energy to excite a material at a lower frequency than it does at a higher one. Higher frequencies give higher fidelity, but are less resistant to signal loss.
I generally use Visio for this sort of thing, but planning out a large network using an automated software tool is impossible, at least in my experience. If you can logically group your network at some level, you will find that it will make planning, and/or diagramming your network much easier. Even the largest networks are broken down into several groups at some level, and you have to keep those groups in mind. You should know the usage patterns of these groups, how much bandwidth each group needs, and where they connect to, etc. Logically map out the best performance scenario for these groups. Then start with your smallest groups, and find similarities to form larger groups, until you have encompassed the entire network. I'm assuming that there are probably departmental connections needed, as well as outside connections, but the key is how much each group of people use each resource available.
Many "well-planned" networks fail because they aimed to provide all resources evenly, but in the real-world, this is not the case. Most of the time, a certain group of people use specific resources, and use very little of other resources. Also keep in mind that you must be able to adapt if the resource requirements for a group change, because it will happen. It's the same thing as trying to use a database wizard to optimize your database. A computer just can't do that for you. The software doesn't have all the necessary input parameters to the problem, and even if it did, the problem is NP Hard. So, in reality, it doesn't matter what tool you use to diagram the network, but actual design still has to be done by human insight, or at least a well-trained monkey...
Ok first of all, regardless of who you ship with, get insurance, and make sure that the person receiving the package has to sign for it, preferably while providing ID. Airborne in my experience has been the best here, as the UPS driver just leaves any package in front of the door, rings the doorbell, and runs back to his truck. If you run to the door, you may catch him leaving, but if you aren't home, anyone could pick up your package.
Either way, you should make sure you are there to pick up the package when it arrives, check the box requiring signature, insure it, and if convenient, tell the shipper to hold your item for pickup at their location. If they lose it at their location, you win. If you didn't sign for it, you win. All shippers have some bad drivers/handlers, and even if you know the particular driver for the destination, you should ALWAYS protect yourself from theft, damage, or other problems. Everything will probably be fine, but when it isn't, those who were the most cautious get their stuff back, and those who said "ship this here" will have a hard time getting anything back.
Maybe that's what michael was referring to! Or better yet, a broken piece of 0.3mm pencil lead. Quick! To the patent office!
Yeah, but if the flight simulator crashes as well as his web server, we don't really need to worry...
Slashdot: Turning expensive hardware into smoking goo since...
Another interesting point. I was reading a story on Yahoo about converting to Linux from Windows. Ernie Ball (a company that primarily makes music hardware/parts/guitar strings) was ordered to pay $90,000 to Microsoft within 120 days to be compliant (not enough licenses). The company used the court ordered 120 day compliance period to evaluate and install Linux, and from the article, never paid a dime.
The same thing could easily happen in this case. IF there is SCO IP in Linux, whatever code it is can be changed in minutes or hours, and even IF SCO was right, they might not get anything! Of course the other good example is USL vs. BSDI, but SCO apparently never paid attention in their history or business law classes...
Evidence? Where? So far, SCO and (maybe) IBM notwithstanding, I believe only a few people have seen ANY evidence in this case.
I think what you meant is "this seems like just as much evidence as SCO has presented", which is true. This equation should help. Let S be all evidence SCO has presented for public viewing. Let P be all public evidence presented against SCO. S = P = lim x->infinity (1/x).
No, a computer language is a tool. Choice of language is a combination of finding which tool is best suited to the task at hand, and personal preference. It's the same choice you'd encounter trying to drive a nail into a board. Writing directly in machine language is like driving the nail in with your forehead, ASM is like using the handle of a screwdriver, and pretty much every other language is like using a hammer. It really doesn't matter what brand of hammer you use, they all will do the task for you, although with varying degrees of effort.
The question is, if you chose hammer x over hammer y to drive in your nail, but you miss the nail and split open your thumb, is it the hammer's fault? Bad coding is like hitting yourself with a hammer, you've got nobody to blame but yourself, but everyone always yells "stupid f&%$ing hammer!"
Not exactly. Assuming that the hardware is ok, you can prove that a system is reliable for any given finite input (including, most importantly, all possible finite substrings of inputs, however it is not possible to test all possible inputs, since a portion of those are infinite), it's just that doing so in large systems takes enormous amounts of time, and of course, time = money. Take Microsoft, for example. It takes a team years to develop a product like Windows XP, run a few test cases, and fix the major bugs. But just think how long it would take to go through every possible input substring of a given length (and by substring/string I am including non-character inputs [mouse, network, etc]).
Consider a simple program that inputs 10 short strings of text and does some computations on those strings. Say for example that the system that has only a keyboard as input, that all input functions are guaranteed only to input A-Z (caps only), the space bar, and 0-9 (regex ((A-Z)*(0-9)*)*( )*), not to overflow, and that there are 10 inputs with exactly 10 characters for each input (spaces fill end of string). This means that there are 37 possibilities for each digit, totaling 37^100 unique possible inputs, about 6.61E156 possibilities, each 100 characters. Typing a million characters per second would take 2.094E145 years! Keep in mind that this is an extremely simple system.
Therefore, it is not possible to test ALL input cases of any nontrivial program, only a few selected cases, which most will agree is far from proving a program correct. Instead, developers should have detailed mathematical descriptions of how a program is to behave at each incremental step, and verify that the program follows those descriptions accurately. Programs can only be proven correct in the same manner that any discrete mathematic concept can be proven correct, with one of the most common methods of a functionality proof being mathematical induction. Based on a few basic assumptions (like that the functions you call work as documented), the rest of the system can be proven by proving the trivial parts and cases first, and then constructing a complete proof based on the trivial parts.
The problem with this is that a small change can have a big impact on the proof, and nobody actually takes the time to verify that everything still works. Companies don't often spend money on making their software 100% correct, they just need to add the nifty new features that their customers want before their competitors do. I'd be willing to bet that 90% of the bugs found in XP can be traced to a "nifty new feature" that broke code that may have been proven correct at some point.
In other words, the short answer is yes, if you can test every state, you can prove a program correct, but since that's usually impossible, it becomes the developers' responsibility to incrementally prove the system, which is far easier if all functionality is planned ahead of time, but still too time/money consuming for most software companies to bother with. Microsoft doesn't care if your computer crashes, you'll probably still pay them, and as much as I'd like to think otherwise, OSS isn't much different (although it's usually more time than money there).
It is nice to have a big standard library, but you also have to consider what is included and what isn't. What often happens (MIL STD 1812 - The Ada Language) is that you get a HUGE library of functions, yet someone can still come up with a few routines that 50% of your programs will use, but aren't included in the library. Also, you may only use 3 or 4 of the library functions, out of the thousands available in such a language. This wastes space.
Also, at some point you may need to control not only what your code does, but how it does it. Ever written something for an embedded system or special hardware in Java, Perl, or Python? Didn't think so. C is as popular as it is for two reasons:
1) Nice syntactic properties, simple language (which is probably why the majority of other languages use a syntax like C, which of course is actually derived from ALGOL).
and 2) It allows very low level access to the machine, thus letting programmers have something to read which can easily be translated to assembly or object code (I believe all keywords and operators map to a single opcode on most architectures, excluding load/store operations and accumulator machines).
In other words, if you don't like C/C++, don't do anything that requires control over what happens. Hell, while you're at it, why not just use Visual Basic to write all your apps, after all, we all know the Microsoft libraries are (HEAVY)(SARCASM)flawless.(/SARCASM)(/HEAVY)
I agree, though it doesn't necessarily need to cover the differences, just maybe tout a few of the traits of the BSD's and OpenBSD specifically. One easy example is the secuity issue. I also think some history from at least 4.4BSD should be covered, since "The Design and Implementation of the 4.4BSD Operating System" is an excellent book and covers the earlier history. Definitely should point out that the OpenBSD camp leans heavily towards security and stability in general.
The book should also cover topics on the basics, configuring services to be run at init, but should avoid trying to be a "Complete Reference", since no book ever really is. Just some basic administrative tasks, things that any sysadmin would do on a regular basis (configuring, building, and testing a new kernel), and the things none of us ever really want to have to do (Recovery of bad slices). Hardware configuration should also be covered, as that is certainly one of the more difficult things for the Win95 and up generation. (What, you mean I actually have to know what I have inside that shiny metal box thingy?) Leave the specifics of major services (DNS, Web, *mail, etc...) up to the books written for them, just overview the services, and give sysadmins a good reference.
That reminds me... If anyone has ever read the Coherent (circa 1994) manuals, those were excellent, and I wish there were something like that now. Complete command references, system architecture overviews, good stuff that is useful to admins and newbies.
This is not a sig. (No really, it's not)
The software errors/stupidity that let the alleged criminals find out they were being tapped, or the fact that we don't have those errors here in the US. Not that I do anything wrong, but I'd still prefer not to be monitored by the government. This reminds me of a /. article a while back about the US taking bids for a central repository of personal information. It may sound like a great idea to the politicians, but after a while they're going to need money, and guess what?... Then everyone's personal information is up for sale. Telcos have been doing this for a while, even my university (UCF) does this, and I get a few dozen porn/marketing spams a day, just because they have my email address in their records.
And you thought spam was a problem now....
Yeah, this is great. Now we'll have MORE stories from Microsoft employees about how great their lives are now that they've switched to Frontpage and Passport. Who knows, maybe this time they won't use a stock photo, too!
Microsoft: Un*x administrator switches to Passport, says 'much more reliable'
AP (next day): Microsoft 'Un*x administrator' found, works for Microsoft PR firm... sound familiar?