With a securelevel of two the kernel cannot be altered and even root cannot fiddle with disks and memory devices. Your are pretty sure your kernel is yours unless the machine is rebooted, but you could notice that. (Why the hell have I been disconnected ?)
Don't be so sure about noticing reboot.
I think it's theoretically possible to reboot a system, replace the kernel, and bring back all of the processes as if nothing has happened. This of course requires grabbing images of the processes and their state before rebooting. Really, if you can replace the kernel, you can do practically anything.
Heck, on most servers it would probably be sufficient to reboot the machine in the middle of the night and then fiddle with the system uptime and processes' run time to make it look like the uptime was still 200+ days.
So, when using securelevels it is important to make sure that everything involved in the boot process is locked down.
Yes, it sent chills down my spine when I read it as well. I've known such things were possible but didn't think anyone had yet gone to the trouble.
There are things you can do about it, though.
Some Unixes, including Linux and the freeware BSDs (all BSDs since 4.4, I think), have the concept of "securelevels". Set files to be immutable (under *BSD the command is "chflags schg somefile") and raise the securelevel above zero. This prevents everyone, including root, from modifying the file. At securelevel 2, the disk and memory devices are also read-only, to prevent doctoring that way.
This doesn't stop intruders from gaining root, but it can prevent them from trojaning everything and going invisible, or at least make it a hell of a lot harder.
The only way around it is to go to the console and bring the system to single-user mode. If some files or directories used in the boot sequence before the securelevel is raised aren't set immutable, it's often possible to modify them such that the securelevel will not be raised during the next reboot, so it's important to know what you're doing. Other than that, the only way for an intruder to trojan the system is to discover a bug in the kernel itself. There have been bugs found in the past, but they are much less plentiful than root exploits.
And ths BIG stink in my eyes is tha fact the NIST eliminated stronger contestants. HPC and CAST-256 have no known weeknesses. MARS, RC6 and TWOFISH all have weekneses!!!!!!! That's right. Read this again. Attckes have been shown to work for them. Not break them wide open mind you, just it's not 2^128 or 2^256 possibilities anymore.
Read the report. HPC does have a serious weakness (equivalent keys, IIRC). And CAST-256 was eliminated because of it's mediocre performance.
Mars, RC6 and Twofish have NOT had any real weaknesses discovered. Any "weeknesses" are really just interesting observations, and can't be used to reduce the workfactor. It is still 2**128 or 2**256 (or 2**192, or other) possibilities.
Of the five that made it to round 2, Mars and RC6 can probably be counted out right away. Mars is too complicated and RC6 doesn't have a large security margin. And both are highly platform-dependant for their speed.
Serpent (one of the non-US ones) will probably be counted out because of it's slow speed, although the high security margin might still save it. One could argue that as CPUs get faster speed becomes a non-issue compared to security. Just look at the popularity of Triple-DES even today.
Rijndael (the other non-US one) and Twofish appear to be the favorites. The report listed no real complaints about Rijndael. Twofish is kinda complicated, but has some space/time tradeoff options that might be worth it for low-memory systems.
Rijndael has a structure that can be parallelized. This could be a very good thing if processing goes that way. Considering that AES is expected to serve for decades, performance on future processors could be very important, though entirely speculative.
Just don't hold your breath. It'll probably be years before we see a winner.
Lots of computer geeks have been more influential than Linus Torvalds. Dennis Ritchie is an obvious one (The C programming language, and Unix). So is Donald Knuth ("The Art of Computer Programming"). Even Bill Joy (BSD, and Sun too I think?) was more influential than Linus.
I had to vote for Alan Turing though. Not only did he invent the model of computing, he also broke Enigma. If not for that WWII might have gone the other way.
And of course, his model of computing has affected the space race, the arms race, etc. Not as significant as mass production, but hey, he was "one of us".
What is the difference between a $400 PC and a $200 Dreamcast? Not much. The PSII will blur the lines even more, and the next generation Nintendo will erase the lines altogether.
Give that Anonymous Coward a cigar!:)
Game consoles have been tremendously sucessful at penetrating the home market, especially compared with the PC. That's because of the low cost and the ease-of-use you get when there's zero configurability. They are also marketed to a single task: Playing games.
Most (l)users don't understand the concept of a general-purpose device. Describe a Turing machine to most people, then tell them that on a fundamental level that's all a computer is, and they'll look at you like you're nuts. Even telling people that Nintendos and PCs are based on the same technology is likely to get a few blank stares.
Sell something as an appliance and people will pay hundreds of dollars for it, even though it only performs one task. Try to sell someone one of many similar devices, with prices all over the four-figure range, none of which will do anything unless you buy more stuff (software) to go with it, and you'll confuse the hell out of people.
It's the question of the killer app. For PCs it was the spreadsheet, followed by other office programs. For game consoles it was, well, games. For modems it was the internet.
If you can sell someone the killer app, without also trying to sell them a general-purpose device, you'll have a much easier time marketing it.
Which computer do you think the average person wants to buy...
Or this one: Word Processor, Spreadsheet, Desktop publishing, Games, Internet
The first is just technobabble. The second is something they can actually use. The fact that they may in fact be the same machine (or not) is beside the point.
If the big companies start integrating everything, and selling the applications, especially "Internet", people will be more likely to buy them than some part list.
"It's what BigCo is selling, so it must be a good computer! It does everything we need. And it only cost us $399!"
The idea of using Celerons only works if you can overclock them. It may go well for hacks and tinkers, but for the heavy video and I/O, a 66Mhz bus frequency is unacceptable.
In the article, the guy ran one of the chips at 2.1 volts... I wonder how long it stayed stable at that speed/voltage. I have a C300A that runs perfectly stable at 450 MHz 1.9v, but crashes regularly if I use default 2.0 voltage. These PPGA Celerons don't dissipate heat as well as their Slot-1 brothers can.
I would expect memory bandwidth to be an issue with SMP Celeron systems, especially if you're not overclocking. The last part of the article touched on this, but I think a lot of people still don't get it...
The Celerons only have 128k cache, compared to 512k on the P2. The Celeron cache runs at twice the speed of the P2 cache, often resulting in better single-processor performance, but because the Celeron cache is smaller it has to go to RAM more often. This means that the Celerons consume more memory bandwith than the P2s.
Combine this with the fact that an un-OCed Celeron runs with only a 66 MHz bus, compared with 100 MHz for a P2. Even though the Celeron needs more memory bandwidth than a P2, it gets less.
Now stick two of these bandwidth-hungry chips on the same board accessing the same 66 MHz RAM...
I suspect it would not be worth it unless you used 300As at 450+ (100+ MHz FSB).
Does anyone have performance numbers for un-overclocked Celeron SMP?
What little that I understand about the quantum microverse and quantum computing, leads me to think that anything quantum can *do*, it can *undo*. Although you may be able to encrypt huge keys, quantum allows you to try ALL the combinations on the lock at once, no matter how large.
I don't know much about quantum computing either, but from what I've read in the context of cryptography, quantum computers are able to solve many (most? all?) problems using the square root of the number of operations needed by conventional computers.
This would make a 128-bit key solvable in 2^64 operations. You would need a 256-bit key to have the same security as a 128-bit key seems to offer today.
Quantum computers will not be able to test every possibility simultaneously. I don't recall details, but I think it had something to do with getting only a certain amount of information about the answer out of each measurement.
Oh boy, there's some fun: first, you have to build your own communication medium, because phone lines and satellites are out. It has to be strong, it has to be extremely accurate, and it has to be completely isolated. Oh, and you'll need to make many different versions. And they all have to be hard-wired, as you can't do this sort of thing over radio waves.
Benchmarking using a real network would make sense if you were benchmarking an entire system, including the network cards and drivers. This didn't seem to be a "this server configuration vs. that server configuration", but rather a "FreeBSD vs. Linux" and "static vs. mod_perl vs. CGI" benchmark.
If you want to see a FreeBSD vs. Linux benchmark, using a real network will skew the results in favour of whichever has the better driver for the given card.
For example, if you benchmarked with an Intel Etherexpress 10/100(B|+) card, the results would probably be tilted in FreeBSD's favour, as that driver in FreeBSD is extremely clean and efficient. OTOH, if you benchmarked with a DEC Tulip-based card, the results would be tilted towards Linux, because the Tulip is maintained much more actively in Linux than in FreeBSD. I'm not talking about a guaranteed win for one OS or the other, but it would have an effect on the result.
This doesn't invalidate your concerns about running a benchmark on localhost. The loopback device in one OS still might be better optimized in one OS, but using a real network device has that problem too.
The client load certainly interferes with testing the system as if it were a real web server. But if it's just some arbitrary FreeBSD vs. Linux test, there's nothing wrong with adding client load, as long as it's the same load on both systems. Such a test is valid, but not interesting, as it doesn't represent a real-world application.
I'd say the "Static vs. mod_perl vs. CGI binary vs. CGI perl script" data is interesting, as all other factors were the same between those tests. I certainly found it educational... I knew what order they would come in, but was not sure exactly what the gaps would be.
The answer is simple: Not all user needs are the same.
Prime examples of why Linux is better for newbies: The basic tasks of installing 3rd party software, and kernel configuration...
For a new user, trying to figure out what the options in a BSD kernel config file do is hellish. The menu-based system Linux has is easy to figure out and much more comfortable for newbies, especially if they come from a Windows background where everything is menu-based. On the other hand, if I (a FreeBSD user of about 4 years) just want to make a quick change, firing up a text editor, finding the line I want to edit, changing it, and saving the file can take well under 5 seconds. Digging through a menu system is a bit more tedious. While it is possible to cross the systems (have a menu program that creates a text config, or a text config that sets the menu options) there doesn't seem to be sufficient demand in either camp AFAICS (though I haven't used Linux since Debian 1.1). Both camps are happy with their way of doing things.
Third-party software install is also different. Under Linux, there are fancy menu-based (usually also accessable by command-line) programs that are devoted to package management. Again, new users will find a menu system to be more comfortable. On the other hand, *BSD, instead of using a special menu-based packaging program, uses a "ports" system that is based on standard compiler tools. You cd to the software's skelton port directory (which contains only a few files, and the entire ports tree is just 8 megs to download) and type "make install". The ports system automatically does dependancy checking and installation, downloads the actual software from one of the distribution sites, checks the MD5 signature, unpacks the files, applies any patches that might be needed, compiles the software, installs it, and registers the installation. It does everything the fancy packaging software in Linux does, but does it the "Unix way" of combining a bunch of generalized tools to perform a specific task.
The important thing to remember though, is that both systems work. The differences I've outlined above (which are probably the two most obvious differences for most users) are religious differences. Both systems work but you get to choose which system you prefer. New users almost always prefer the Linux way (which is why BSDers consider Linux a "training ground"), but once they learn how things work they actually have a choice in what religion they prefer for their everyday activities.
Can you not put the system OS disk and one of the striped disks on the primary controller, and the other striped disk on the secondary?
The idea is to split things up so that disks you are likely to be accessing at the same time will be on seperate controllers. The system disk probably doesn't matter as much as the stripped disks, because the frequently accessed stuff on the system disk tends to stay in RAM anyway.
By the way, do you have to do anything special to keep your system cool with three disks? I had a 1 GB and a 340 MB disk in my system not long ago, and they warmed the inside of the whole box, even with two fans. My single 13 GB is cooler. (I guess 5400 RPM produces less heat than 3600 RPM + 3600 RPM)
You might try striping (RAID-0) instead. It offers performance advantages like real RAID but does not provide redundancy. It only requires two disks, though more disks gives better concurrency (unless they're on the same IDE chain!).
It's less reliable than real RAID for sure, but it's also even less reliable than using seperate disks as seperate disks. If one fails you effectively lose the contents of both.
Still, for an experimental box, or a box where you care more about performance than the data safety (USENET server), it's cheaper than RAID-5.
In practical terms, how is the BSD licence (without advertising clause) different from releasing something to the public domain?
I know there are significant legal differences, but I'm wondering how that translates into practical differences.
As far as I can tell, either way, other people can use your code however they want without telling you. You can also still use the code however you want: under BSD you still own it, under PD everyone owns it. Right?
With PD other people can re-release the code under a different licence, under BSD they can't, but either way they can still use the code however they want. So practically speaking, what is the difference?
Consiousness is the process of updating your internal representation of yourself, and that's all.
You just blew my mind. Er, wait... Maybe you just updated my internal representation of my internal representation of myself. Or something like that.
Your AI background, email address, and your choice of words suggests to me that you probably have a file named conciousness.c in your home directory. Care to release it under an open-source license?:)
Anyway, to put your statement into less programmer-like words, it sounds like you're saying that conciousness is our awareness of our own awareness. In a recursive-like fashion, we are also aware of our own awareness of our own awareness, etc. Maybe this is an interesting observation, sort of like how a turing machine can implement a turing machine?
That's not to say that we have to be fully aware of our own awareness in order to be concious. Lots of stuff goes on in our own minds that we are totally unaware of: specifically, the subconcious. "Look, Ma! I'm aware of my own lack of awareness!".;)
All this thinking involving conciousness, AI, recursion, and turing machines has lead me to the question:
How many thoughts do you think a thought thought could think if a thought thought could think thoughts?
This gives me a headache. However, I am aware of the headache, and am also aware of my own awareness of it. As you said, my internal representation of myself has been updated. Or has it? How, exactly, do you even have an internal representation of yourself? And how is it being updated? And how can this internal representation of yourself produce such an idea as the "internal representation of yourself"? If we understand conciousness, do we really understand it, or do we just think we understand it???
I thought a thought. Of thinking I thought. I thought a thought of thinking. On thinking this thought I raised my thought Of thinking, Or so I thought.
Don't you folks get it? All ideas are memes. Religious beliefs certainly are memes. The whole idea of "memes" is also a meme. Even the scientific method is a meme.
These are all ideas that have survived and flourished in people's minds, and continue to perpetuate themselves in one way or another.
The "Good Times virus" and similar hoaxes are also memes, and are obvious examples of how memes can spread like viruses.
It used to be fun to count how long it would take him/her/it after a new wterm was released to change all the credits again and release a new aterm, also you could appear like an oracle to newbies by predicting an aterm release and what it's new features would be.
Um, Are you sure about that? I'm not familiar with the history of the two, but the reason I picked aterm over wterm is that aterm has features wterm lacked/lacks. Specifically, shading rather than just tinting with primary colours. I tried aterm and wterm at the same time to compare, and aterm could shade arbitrary colours without turning the background into a four-bit mess. How could aterm "rip off" from wterm what wterm didn't/doesn't have?
In fact the whole afterstep camp seem to suffer from the same problem - following in the footsteps of Window Maker.
Once all you can do is rip off everyone else, your project and community is dead.
As I understand it, Afterstep was thriving before Window Maker came along. I don't claim to know what happened after Window Maker appeared or what code has been "ripped off", but the fact that Afterstep has some features (transparent title dock and menus, as pointed out by the previous poster) suggests that they have been doing some interesting stuff and not just "following in the footsteps of Window Maker".
I also use shaded aterms, but with WindowMaker. Aterm development seems to be farther along than wterm. I hate to say it, but I think the wterm folks are just duplicating effort.:(
I tried eterm too, but it seemed bloated and unstable. That was a while back, so maybe it's improved, but aterm does everthing I want so I haven't bothered to re-evaluate eterm.
I noticed in your screenshot that you aren't using the transparent scrollbars. I know they are in aterm 0.3.6, as I use them myself, and they look much nicer than the plain gray ones. Stick this in your ~/.Xdefaults file:
Aterm*transpscrollbar: true
Transparent dock and menus sounds cool. If titlebars were also transparent you would never need to pick a theme, just a background.
Now if I could just get every other X application to do transparency...:)
> MP3 *is* a lot more compact than CD audio is. > Does that mean CD audio, being 50MB in size, > will also die because we don't have the > bandwidth? It sure doesn't.
It sure does.:) Think about it...
The lack of compression in CD tracks is why MP3s exist. You could dump a raw CD track into a file, but why? MP3s are much faster to upload/download.
CDs are artifacts of the unconnected days. The only reason they still exist are because most people aren't wired to modern CPUs all of the time.
Slashdot Mirror
How the universe will end is, I think, still an open question.
If the universe is "closed", it will end at some point with a big crunch as everything collapses together due to gravity.
If the universe is "open", it will continue expanding forever.
If I remember correctly, current thinking is that the universe is "open".
Disclaimer: I am not a physicist.
Don't be so sure about noticing reboot.
I think it's theoretically possible to reboot a system, replace the kernel, and bring back all of the processes as if nothing has happened. This of course requires grabbing images of the processes and their state before rebooting. Really, if you can replace the kernel, you can do practically anything.
Heck, on most servers it would probably be sufficient to reboot the machine in the middle of the night and then fiddle with the system uptime and processes' run time to make it look like the uptime was still 200+ days.
So, when using securelevels it is important to make sure that everything involved in the boot process is locked down.
Yes, it sent chills down my spine when I read it as well. I've known such things were possible but didn't think anyone had yet gone to the trouble.
There are things you can do about it, though.
Some Unixes, including Linux and the freeware BSDs (all BSDs since 4.4, I think), have the concept of "securelevels". Set files to be immutable (under *BSD the command is "chflags schg somefile") and raise the securelevel above zero. This prevents everyone, including root, from modifying the file. At securelevel 2, the disk and memory devices are also read-only, to prevent doctoring that way.
This doesn't stop intruders from gaining root, but it can prevent them from trojaning everything and going invisible, or at least make it a hell of a lot harder.
The only way around it is to go to the console and bring the system to single-user mode. If some files or directories used in the boot sequence before the securelevel is raised aren't set immutable, it's often possible to modify them such that the securelevel will not be raised during the next reboot, so it's important to know what you're doing. Other than that, the only way for an intruder to trojan the system is to discover a bug in the kernel itself. There have been bugs found in the past, but they are much less plentiful than root exploits.
Read the report. HPC does have a serious weakness (equivalent keys, IIRC). And CAST-256 was eliminated because of it's mediocre performance.
Mars, RC6 and Twofish have NOT had any real weaknesses discovered. Any "weeknesses" are really just interesting observations, and can't be used to reduce the workfactor. It is still 2**128 or 2**256 (or 2**192, or other) possibilities.
Yeah, Rijndael appears to have a good chance at becoming the AES.
Check out NIST's Round 1 Report (PDF) for the raw details if you haven't already.
Of the five that made it to round 2, Mars and RC6 can probably be counted out right away. Mars is too complicated and RC6 doesn't have a large security margin. And both are highly platform-dependant for their speed.
Serpent (one of the non-US ones) will probably be counted out because of it's slow speed, although the high security margin might still save it. One could argue that as CPUs get faster speed becomes a non-issue compared to security. Just look at the popularity of Triple-DES even today.
Rijndael (the other non-US one) and Twofish appear to be the favorites. The report listed no real complaints about Rijndael. Twofish is kinda complicated, but has some space/time tradeoff options that might be worth it for low-memory systems.
Rijndael has a structure that can be parallelized. This could be a very good thing if processing goes that way. Considering that AES is expected to serve for decades, performance on future processors could be very important, though entirely speculative.
Just don't hold your breath. It'll probably be years before we see a winner.
Looks like they just garbled their own english. 128, 192 and 256 bit are keysizes required for AES.
What really bugged me is the "340**35" number at the bottom. It looks like someone just pulled some random base and exponent out of thin air.
Most reporters take pride in their accuracy. *snicker* Oh well, I guess reporters get confused by technical stuff just like all other non-techies.
Lots of computer geeks have been more influential than Linus Torvalds. Dennis Ritchie is an obvious one (The C programming language, and Unix). So is Donald Knuth ("The Art of Computer Programming"). Even Bill Joy (BSD, and Sun too I think?) was more influential than Linus.
I had to vote for Alan Turing though. Not only did he invent the model of computing, he also broke Enigma. If not for that WWII might have gone the other way.
And of course, his model of computing has affected the space race, the arms race, etc. Not as significant as mass production, but hey, he was "one of us".
What is the difference between a $400 PC and a $200 Dreamcast? Not much. The PSII will blur the lines even more, and the next generation Nintendo will erase the lines altogether.
Give that Anonymous Coward a cigar! :)
Game consoles have been tremendously sucessful at penetrating the home market, especially compared with the PC. That's because of the low cost and the ease-of-use you get when there's zero configurability. They are also marketed to a single task: Playing games.
Most (l)users don't understand the concept of a general-purpose device. Describe a Turing machine to most people, then tell them that on a fundamental level that's all a computer is, and they'll look at you like you're nuts. Even telling people that Nintendos and PCs are based on the same technology is likely to get a few blank stares.
Sell something as an appliance and people will pay hundreds of dollars for it, even though it only performs one task. Try to sell someone one of many similar devices, with prices all over the four-figure range, none of which will do anything unless you buy more stuff (software) to go with it, and you'll confuse the hell out of people.
It's the question of the killer app. For PCs it was the spreadsheet, followed by other office programs. For game consoles it was, well, games. For modems it was the internet.
If you can sell someone the killer app, without also trying to sell them a general-purpose device, you'll have a much easier time marketing it.
Which computer do you think the average person wants to buy...
This one:
PIII-550 MHz, 128MB RAM, 17 GB HD, TNT2 Ultra, 56k Modem
Or this one:
Word Processor, Spreadsheet, Desktop publishing, Games, Internet
The first is just technobabble. The second is something they can actually use. The fact that they may in fact be the same machine (or not) is beside the point.
If the big companies start integrating everything, and selling the applications, especially "Internet", people will be more likely to buy them than some part list.
"It's what BigCo is selling, so it must be a good computer! It does everything we need. And it only cost us $399!"
Cool, but how does it SOUND? All the computer speach products I've heard sound very monotonous.
The idea of using Celerons only works if you can overclock them. It may go well for hacks and tinkers, but for the heavy video and I/O, a 66Mhz bus frequency is unacceptable.
In the article, the guy ran one of the chips at 2.1 volts... I wonder how long it stayed stable at that speed/voltage. I have a C300A that runs perfectly stable at 450 MHz 1.9v, but crashes regularly if I use default 2.0 voltage. These PPGA Celerons don't dissipate heat as well as their Slot-1 brothers can.
I would expect memory bandwidth to be an issue with SMP Celeron systems, especially if you're not overclocking. The last part of the article touched on this, but I think a lot of people still don't get it...
The Celerons only have 128k cache, compared to 512k on the P2. The Celeron cache runs at twice the speed of the P2 cache, often resulting in better single-processor performance, but because the Celeron cache is smaller it has to go to RAM more often. This means that the Celerons consume more memory bandwith than the P2s.
Combine this with the fact that an un-OCed Celeron runs with only a 66 MHz bus, compared with 100 MHz for a P2. Even though the Celeron needs more memory bandwidth than a P2, it gets less.
Now stick two of these bandwidth-hungry chips on the same board accessing the same 66 MHz RAM...
I suspect it would not be worth it unless you used 300As at 450+ (100+ MHz FSB).
Does anyone have performance numbers for un-overclocked Celeron SMP?
When iris-scanning ATM's arrive in my neighborhood, they'll be trashed inside of 24 hours.
More likely, the users will be trashed...
Machine: Welcome to Personal Iris Banking. To begin, lean forward and look directly into the scanner.
You lean forward...
Mugger in the shadows: Yeah, lean forward sucker.
Machine: Scan not completed. Please lean forward and look directly into the scanner.
You: Damn machine...
You lean a bit farther forward to get a proper scan.
The mugger seizes the opportunity and sneaks up behind you.
*** THWAP! ***
You: Ungh!
Mugger: Shuddup!
Machine: You currently have $1,234.56 in your account. Would you like to make a withdrawl?
You: Ungh!
Mugger punches in a withdrawl, takes the money, and runs off.
Machine: You have $0.00 remaining in your account. Thank you for using Personal Iris Banking.
You: Ungh!
Machine: Oh, and happy birthday Mr. Smith.
With quantum computing does that mean we could only view the contents of a Micros~1 Word document or modify them but not both?
No. It means that you can not view it without changing it. ;)
What little that I understand about the quantum microverse and quantum computing, leads me to think that anything quantum can *do*, it can *undo*. Although you may be able to encrypt huge keys, quantum allows you to try ALL the combinations on the lock at once, no matter how large.
I don't know much about quantum computing either, but from what I've read in the context of cryptography, quantum computers are able to solve many (most? all?) problems using the square root of the number of operations needed by conventional computers.
This would make a 128-bit key solvable in 2^64 operations. You would need a 256-bit key to have the same security as a 128-bit key seems to offer today.
Quantum computers will not be able to test every possibility simultaneously. I don't recall details, but I think it had something to do with getting only a certain amount of information about the answer out of each measurement.
Oh boy, there's some fun: first, you have to build your own communication medium, because phone lines and satellites are out. It has to be strong, it has to be extremely accurate, and it has to be completely isolated. Oh, and you'll need to make many different versions. And they all have to be hard-wired, as you can't do this sort of thing over radio waves.
Packet-switched networks are out too (I think?).
:(
Benchmarking using a real network would make sense if you were benchmarking an entire system, including the network cards and drivers. This didn't seem to be a "this server configuration vs. that server configuration", but rather a "FreeBSD vs. Linux" and "static vs. mod_perl vs. CGI" benchmark.
If you want to see a FreeBSD vs. Linux benchmark, using a real network will skew the results in favour of whichever has the better driver for the given card.
For example, if you benchmarked with an Intel Etherexpress 10/100(B|+) card, the results would probably be tilted in FreeBSD's favour, as that driver in FreeBSD is extremely clean and efficient. OTOH, if you benchmarked with a DEC Tulip-based card, the results would be tilted towards Linux, because the Tulip is maintained much more actively in Linux than in FreeBSD. I'm not talking about a guaranteed win for one OS or the other, but it would have an effect on the result.
This doesn't invalidate your concerns about running a benchmark on localhost. The loopback device in one OS still might be better optimized in one OS, but using a real network device has that problem too.
The client load certainly interferes with testing the system as if it were a real web server. But if it's just some arbitrary FreeBSD vs. Linux test, there's nothing wrong with adding client load, as long as it's the same load on both systems. Such a test is valid, but not interesting, as it doesn't represent a real-world application.
I'd say the "Static vs. mod_perl vs. CGI binary vs. CGI perl script" data is interesting, as all other factors were the same between those tests. I certainly found it educational... I knew what order they would come in, but was not sure exactly what the gaps would be.
The answer is simple: Not all user needs are the same.
Prime examples of why Linux is better for newbies: The basic tasks of installing 3rd party software, and kernel configuration...
For a new user, trying to figure out what the options in a BSD kernel config file do is hellish. The menu-based system Linux has is easy to figure out and much more comfortable for newbies, especially if they come from a Windows background where everything is menu-based. On the other hand, if I (a FreeBSD user of about 4 years) just want to make a quick change, firing up a text editor, finding the line I want to edit, changing it, and saving the file can take well under 5 seconds. Digging through a menu system is a bit more tedious. While it is possible to cross the systems (have a menu program that creates a text config, or a text config that sets the menu options) there doesn't seem to be sufficient demand in either camp AFAICS (though I haven't used Linux since Debian 1.1). Both camps are happy with their way of doing things.
Third-party software install is also different. Under Linux, there are fancy menu-based (usually also accessable by command-line) programs that are devoted to package management. Again, new users will find a menu system to be more comfortable. On the other hand, *BSD, instead of using a special menu-based packaging program, uses a "ports" system that is based on standard compiler tools. You cd to the software's skelton port directory (which contains only a few files, and the entire ports tree is just 8 megs to download) and type "make install". The ports system automatically does dependancy checking and installation, downloads the actual software from one of the distribution sites, checks the MD5 signature, unpacks the files, applies any patches that might be needed, compiles the software, installs it, and registers the installation. It does everything the fancy packaging software in Linux does, but does it the "Unix way" of combining a bunch of generalized tools to perform a specific task.
The important thing to remember though, is that both systems work. The differences I've outlined above (which are probably the two most obvious differences for most users) are religious differences. Both systems work but you get to choose which system you prefer. New users almost always prefer the Linux way (which is why BSDers consider Linux a "training ground"), but once they learn how things work they actually have a choice in what religion they prefer for their everyday activities.
Choice is good.
Can you not put the system OS disk and one of the striped disks on the primary controller, and the other striped disk on the secondary?
The idea is to split things up so that disks you are likely to be accessing at the same time will be on seperate controllers. The system disk probably doesn't matter as much as the stripped disks, because the frequently accessed stuff on the system disk tends to stay in RAM anyway.
By the way, do you have to do anything special to keep your system cool with three disks? I had a 1 GB and a 340 MB disk in my system not long ago, and they warmed the inside of the whole box, even with two fans. My single 13 GB is cooler. (I guess 5400 RPM produces less heat than 3600 RPM + 3600 RPM)
You might try striping (RAID-0) instead. It offers performance advantages like real RAID but does not provide redundancy. It only requires two disks, though more disks gives better concurrency (unless they're on the same IDE chain!).
It's less reliable than real RAID for sure, but it's also even less reliable than using seperate disks as seperate disks. If one fails you effectively lose the contents of both.
Still, for an experimental box, or a box where you care more about performance than the data safety (USENET server), it's cheaper than RAID-5.
In practical terms, how is the BSD licence (without advertising clause) different from releasing something to the public domain?
I know there are significant legal differences, but I'm wondering how that translates into practical differences.
As far as I can tell, either way, other people can use your code however they want without telling you. You can also still use the code however you want: under BSD you still own it, under PD everyone owns it. Right?
With PD other people can re-release the code under a different licence, under BSD they can't, but either way they can still use the code however they want. So practically speaking, what is the difference?
I must be missing something here???
Consiousness is the process of updating your internal representation of yourself, and that's all.
You just blew my mind. Er, wait... Maybe you just updated my internal representation of my internal representation of myself. Or something like that.
Your AI background, email address, and your choice of words suggests to me that you probably have a file named conciousness.c in your home directory. Care to release it under an open-source license? :)
Anyway, to put your statement into less programmer-like words, it sounds like you're saying that conciousness is our awareness of our own awareness. In a recursive-like fashion, we are also aware of our own awareness of our own awareness, etc. Maybe this is an interesting observation, sort of like how a turing machine can implement a turing machine?
That's not to say that we have to be fully aware of our own awareness in order to be concious. Lots of stuff goes on in our own minds that we are totally unaware of: specifically, the subconcious. "Look, Ma! I'm aware of my own lack of awareness!". ;)
All this thinking involving conciousness, AI, recursion, and turing machines has lead me to the question:
How many thoughts do you think a thought thought could think if a thought thought could think thoughts?
This gives me a headache. However, I am aware of the headache, and am also aware of my own awareness of it. As you said, my internal representation of myself has been updated. Or has it? How, exactly, do you even have an internal representation of yourself? And how is it being updated? And how can this internal representation of yourself produce such an idea as the "internal representation of yourself"? If we understand conciousness, do we really understand it, or do we just think we understand it???
Don't you folks get it? All ideas are memes. Religious beliefs certainly are memes. The whole idea of "memes" is also a meme. Even the scientific method is a meme.
These are all ideas that have survived and flourished in people's minds, and continue to perpetuate themselves in one way or another.
The "Good Times virus" and similar hoaxes are also memes, and are obvious examples of how memes can spread like viruses.
It used to be fun to count how long it would take him/her/it after a new wterm was released to change all the credits again and release a new aterm, also you could appear like an oracle to newbies by predicting an aterm release and what it's new features would be.
Um, Are you sure about that? I'm not familiar with the history of the two, but the reason I picked aterm over wterm is that aterm has features wterm lacked/lacks. Specifically, shading rather than just tinting with primary colours. I tried aterm and wterm at the same time to compare, and aterm could shade arbitrary colours without turning the background into a four-bit mess. How could aterm "rip off" from wterm what wterm didn't/doesn't have?
In fact the whole afterstep camp seem to suffer from the same problem - following in the footsteps of Window Maker.
Once all you can do is rip off everyone else, your project and community is dead.
As I understand it, Afterstep was thriving before Window Maker came along. I don't claim to know what happened after Window Maker appeared or what code has been "ripped off", but the fact that Afterstep has some features (transparent title dock and menus, as pointed out by the previous poster) suggests that they have been doing some interesting stuff and not just "following in the footsteps of Window Maker".
Afterstep is dead.
Can't we all just get along?
Admiral Burrito - a happy Window Maker user.
I tried eterm too, but it seemed bloated and unstable. That was a while back, so maybe it's improved, but aterm does everthing I want so I haven't bothered to re-evaluate eterm.
I noticed in your screenshot that you aren't using the transparent scrollbars. I know they are in aterm 0.3.6, as I use them myself, and they look much nicer than the plain gray ones. Stick this in your ~/.Xdefaults file:
Aterm*transpscrollbar: true
Transparent dock and menus sounds cool. If titlebars were also transparent you would never need to pick a theme, just a background.
Now if I could just get every other X application to do transparency... :)
> MP3 *is* a lot more compact than CD audio is.
:) Think about it...
> Does that mean CD audio, being 50MB in size,
> will also die because we don't have the
> bandwidth? It sure doesn't.
It sure does.
The lack of compression in CD tracks is why MP3s
exist. You could dump a raw CD track into a file,
but why? MP3s are much faster to upload/download.
CDs are artifacts of the unconnected days. The
only reason they still exist are because most
people aren't wired to modern CPUs all of the
time.
Am I right?
FOR SALE: F-117A Stealth Fighter/Bomber
Made in the USA! Needs work. $50 OBO
Call Yugoslavia, ask for Slobodan.
P.S. Will accept SAMs in trade.