Part of the problem seems to be that people running mail servers and tools like SpamAssassin aren't running their own caching nameservers locally.
I actually run two caching nameservers. One for email servers and the other for everything else. The DNS lookups for email have a different *context* than web browsing -- which indicates to me that they should be in separate caches. Negative responses for an email lookup shouldn't pollute the positive information about web lookups.
::shrugs:: It's hard to say if this is a case of "brain dead implementations of DNS and mail servers". The biggest problem with an email/DNS setup is that there are so many possibilities and complexities that it's hard to make generalized statements about how site deals with things over another.
I was looking at the parent post and thinking to myself "Goeddles? Goeddles? Who is this guy? Could it be Gödel? Nah, dude said this guy was a "computer science theoretician"; if he meant Gödel he would have said mathematician. Shit, I'm gonna have to google for this Goeddles guy -- I wonder what other cool stuff he said".
im hoping my system starts acting up soon so i finally have an excuse to use itp> Give me an account on your machine, and I'll take care of that for you.
Yeah, I lived about off of Coit and Campbell by the Central Market in Plano. So a few minutes to the east was a DART train station. A few minutes to the west was a DART bus hub. Shuttles or "short routes" from the bus hub to the train? Zero.
Sure, you can take the bus, if you don't mind taking two hours to get somewhere. I lived north of Dallas. There was a small train that went downtown with a stop right across the street from my office. I lived about a 10 minute drive from the station. My thought was, "Hey, I can ride a bus to the station".
Riiight. The bus route from my house to the train station took seven stops and about two hours -- bear in mind that a major "bus hub" was five minutes from my house in the other direction. Nope, no "bus from the bus hub to the train station" in Plano.
It's actually worse than your stats. "Dallas" is a relatively meaningless term. It's a nice label, but you can drive from Dallas to Fort Worth without ever leaving "urban". The metropolitan area is probably closer to 1,200 square miles with a population of 5,000,000. (DFW airport is bigger than the island of Manhattan). The sheer sprawl makes mass transit a much, much more difficult and expensive proposal.
Interesting he talks about this, but don't you usually have seperate support contracts for the OS and your core apps? I have a beast of a box that runs Windows 2000 Advanced Server but I'm free to run any RDBMS or web server I desire on it.
Well, that's partly true and partly untrue. Most support contracts for enterprise applications are very specific about exactly what OS versions you're allowed to run. SAP and Oracle both come to mind: "with version X.Y.Z. of our product, the supported configurations include A.B.C. or A.B.C.D . or D.E.F. but not C.D.E." I've seen the same thing with smaller contact management applications. OS vendors often say things like "well, application X.Y.Y.Z. only runs on OS A.B.C.D. so if you want it to work, you need to apply patch D. Oh, patch D breaks application Y? Sorry.".
OS and application contracts are decoupled in the general case, but are almost always related in specific cases.
...programs can break on subsequent version of Windows becuase something has moved or changed. Come on, trolls. Maybe the reason windows spreads most of the viruses on the internet is because it's so hard to get your PC online without getting 0wn3d.
Lacking a computer science degree, I didn't understand a lot of basic algorithms as well as I should -- some I didn't understand at all. I just used what worked even if something else might have worked better.
Mastering Alorithms with C went a *looong* way toward teaching me the pros and cons of various sorts, trees, graphs and queues. (There's also nice sections on compression and encryption).
The explanations are very thorough. It helps to have at least a minimal understanding of C, but you can get a lot out of the book without it. Lots of pretty pictures and diagrams for the visually inclined.
If you don't have a formal education in computer science, I recommend that you run out and buy this book right now -- you'll be doing yourself a huge favor.
Guide to LaTeX is the "beginner's manual / guide" in the series. (I hesitate to call two books a series, but, hey). I just bought it tonight, and it looks great.
You might want to start off with Guide to LaTeX by Helmut Lopka and Patrick Daly. It's got the same publisher (Addison-Wesley) and is a book about using LaTeX. The LaTeX companion (from what I gathered by perusing it in the bookstore earlier tonight) is more about packages that provide additional functionality to the base system.
I've only had a chance to scan through bits and pieces since I brought it home, but Guide to LaTeX seems to be an excellent book that is a great starting place for learning about LaTeX.
If I had more disposable income I would have bought both of them. I've already seen places in the Guide that refer to the Companion.
I don't think that I would put "Berkeley DB" and "easy data storage" in the same sentence. Real-world use of bdb requires non-trivial application design...you're not going to be able to slap something together like you can with Access. It works great, but it can take some effort.
I've only used the C API, and maybe some of the available wrappers (perl, python) make it more developer friendly. I wouldn't recommend bdb as an answer to this guy's scenario.
I'm a Cedega subscriber. I provide them $$ because I want to support them, not because I actually use the product.
Mind you, if it really worked, I'd use it. I spend a lot of time tweaking, configuring, testing, and researching to get things working just right on my computers. When I sit down to play a game, I want to just put the damn CD in and play it. I don't want to have to do anything else. I want to play a game -- not engage in some research and config fest; I'm trying to play a game because I'm tired of "making things work".
When I do play it, I don't want it to work like some mutant offspring -- "Well, the cinematics don't work" -- "It only works well in 800x600" -- "Go to your X config file and type ${whatever}" -- "The sound? Ahh, it's not *that* important, is it?"
End result? I subscribe to Cedega, and have a big old stack of games that I don't play.
Isnt [sic] this how MS won from Apple and IBM to start with?
I am reminded by a quote from Paul Graham. "But VCs are mistaken to look for the next Microsoft, because no startup can be the next Microsoft unless some other company is prepared to bend over at just the right moment and be the next IBM."
I don't want a UI to provide an "intuitive information landscape" -- I keep the "information landscape" in my head. I want precise and discrete control -- analog versus digital, I suppose. "Do what I say" rather than "do what I mean".
I've been using graphical UIs since the early 80s, and I still get mildly frustrated a few times a day with "must click this, that and the-other" instead of "typetypetype -- done". At least clicking is "digital". I can't imagine trying to use a 3D filesystem UI where my mouse's scroll wheel (or movement or any other "analog" input) is a significant input device. "/var.../var/log/apache/...shit.../var/log...OK, easy now, easy.../var/log/apache/2002/10/httpd.log...FUCK FUCK SHIT!! *SLAM* *SLAM* *SLAM*...nfs:/moogie/home/cvs...AAAARGH!!!"
I will gladly admit that I'm very command-line-centric. The first time I tried to use a Macintosh, I whined "But, but...where are all the files!?". My first home PC was a Kim-1. (Hah! When I was a kid, we didn't *have* keyboards! My idea of intuituve input was wiring switches to the serial port)!
Strange.. I have no such aspirations in other fields. I just like to work on systems and make them do cool stuff.
Ahhh. But what do you define as a system? I, personally, am interested in almost all complex systems -- be it beer-making, groups of people, a person, cooking, computers, agriculture, languages, what-have-you.
I define a complex system as a system with behaviors that I will never be able to 100% accurately predict for any random period of time.
The more control I can exert over J. Random Complex System, the more likely I am to actually tinker with it. Global economics? I can't exert too much control so I'm only interested in a theoretical manner. The beer in the garage; the fields out on the acreage; the computers in the computer room? I can exert a lot of control on those systems so I'm interested in a practical manner.
I'm interested and tinker with a system if I can make inputs into a complex system and claim the output as my own. Systems where I can say, "It worked because I made it," or where I have to say, "I fucked it all up."
Damn! I missed that commit message! Thanks for pointing it out!
It must have come fast on the heels of the following commit message that so enthralled me:
From the changelog:
"1/10/2003: Replaced our TCP/IP stack with one licensed from Microsoft. Work continues on porting over the Linux virtual memory management system. "
Or, if not anxious, you're FUCKING SCARED SHITLESS! (Or bored off of your ass, but that part isn't pertinent to the rest of my conversation).
People that don't experience anxiety or fear in combat are either (a) too stupid to know better...which means they'll probably be dead soon or (b) are probably clinically insane.
Now, don't get me wrong -- combat is the most intense adrenaline rush that you can possibly experience. It's the real-world version of "fight or flight" over and over and over again. Continuously...for hours and days and weeks. And for some people, that adrenaline rush is enjoyable and addictive. But, as an avowed adrenaline junkie, combat is an adrenaline overdose.
There is a military axiom "Train as you intend to fight". Military training is intended to ingrain patterns of behavior that soldiers will fall back on when their capacity for rational thought has been eroded by the stresses of combat. (I will not digress into a discussion of how that training is mostly centered around "how to kill and survive" rather that "how to comply with the Geneva convention and related accords" -- though that might explain why every army has a history of military atrocities).
A computer assistant -- to be effective in combat -- would have to "learn" its user in situations equivalent to combat. If it didn't, it wouldn't react the same in actual combat. And, I assure you, no training simulation comes close to approaching combat. The combination of "If I make a mistake I will die" and "I am actually killing a person" and "I could randomly die no matter what the fuck I do" is not something that can be realistically duplicated outside of the experience. I would also like to point out that soldiers do *not* respond well to "things outside of normal training parameters". The general inclination is to "shoot it" or "evade detection while trying to decide if I should shoot it".
So what we have is (a) a computer program that has "learned" its user in situations less than combat (b) a soldier that has "learned" preset responses to external stimuli -- including the computer. If you throw the soldier+computer into combat, the computer doesn't know how to react anymore -- not correctly anyway -- because it's never "learned" about combat. The soldier will tend to listen to the computer because that's what he's been trained to do...but the computer is going to be offering suboptimal advice/help/whatever...because it doesn't "know" about combat.
There's definitely roles for computers on the battlefield. Trying to provide stress-relief is probably not one of them. Certainly it's an interesting idea. And one day, it might be plausible. But, now, it's not too feasible. I think that computing will have to evolve quite a bit more before battlefield assistants that monitor emotional states and provide feedback based on that emotional state are truly a force multiplier.
Slashdot Mirror
I actually run two caching nameservers. One for email servers and the other for everything else. The DNS lookups for email have a different *context* than web browsing -- which indicates to me that they should be in separate caches. Negative responses for an email lookup shouldn't pollute the positive information about web lookups.
System Shock 2
I was looking at the parent post and thinking to myself "Goeddles? Goeddles? Who is this guy? Could it be Gödel? Nah, dude said this guy was a "computer science theoretician"; if he meant Gödel he would have said mathematician. Shit, I'm gonna have to google for this Goeddles guy -- I wonder what other cool stuff he said".
im hoping my system starts acting up soon so i finally have an excuse to use itp>
Give me an account on your machine, and I'll take care of that for you.
Well, as long as you're not a pederast feel free do be as pedantic as you'd like.
Yeah, I lived about off of Coit and Campbell by the Central Market in Plano. So a few minutes to the east was a DART train station. A few minutes to the west was a DART bus hub. Shuttles or "short routes" from the bus hub to the train? Zero.
What the screaming fuck do "virtual desktops" have to do with "bazillions of itsy-bitsy floating windows", you troll?!
Whew. Glad I got that out of my system.
Sure, you can take the bus, if you don't mind taking two hours to get somewhere.
I lived north of Dallas. There was a small train that went downtown with a stop right across the street from my office. I lived about a 10 minute drive from the station. My thought was, "Hey, I can ride a bus to the station".
Riiight. The bus route from my house to the train station took seven stops and about two hours -- bear in mind that a major "bus hub" was five minutes from my house in the other direction. Nope, no "bus from the bus hub to the train station" in Plano.
It's actually worse than your stats. "Dallas" is a relatively meaningless term. It's a nice label, but you can drive from Dallas to Fort Worth without ever leaving "urban". The metropolitan area is probably closer to 1,200 square miles with a population of 5,000,000. (DFW airport is bigger than the island of Manhattan).
The sheer sprawl makes mass transit a much, much more difficult and expensive proposal.
Interesting he talks about this, but don't you usually have seperate support contracts for the OS and your core apps? I have a beast of a box that runs Windows 2000 Advanced Server but I'm free to run any RDBMS or web server I desire on it.
Well, that's partly true and partly untrue. Most support contracts for enterprise applications are very specific about exactly what OS versions you're allowed to run. SAP and Oracle both come to mind: "with version X.Y.Z. of our product, the supported configurations include A.B.C. or A.B.C.D . or D.E.F. but not C.D.E." I've seen the same thing with smaller contact management applications.
OS vendors often say things like "well, application X.Y.Y.Z. only runs on OS A.B.C.D. so if you want it to work, you need to apply patch D. Oh, patch D breaks application Y? Sorry.".
OS and application contracts are decoupled in the general case, but are almost always related in specific cases.
I nearly spewed Diet Coke out of my nose all over the keyboard. Thanks.
...programs can break on subsequent version of Windows becuase something has moved or changed. Come on, trolls. Maybe the reason windows spreads most of the viruses on the internet is because it's so hard to get your PC online without getting 0wn3d.
Mastering Algorithms with C
Lacking a computer science degree, I didn't understand a lot of basic algorithms as well as I should -- some I didn't understand at all. I just used what worked even if something else might have worked better.
Mastering Alorithms with C went a *looong* way toward teaching me the pros and cons of various sorts, trees, graphs and queues. (There's also nice sections on compression and encryption).
The explanations are very thorough. It helps to have at least a minimal understanding of C, but you can get a lot out of the book without it. Lots of pretty pictures and diagrams for the visually inclined.
If you don't have a formal education in computer science, I recommend that you run out and buy this book right now -- you'll be doing yourself a huge favor.
Guide to LaTeX is the "beginner's manual / guide" in the series. (I hesitate to call two books a series, but, hey). I just bought it tonight, and it looks great.
You might want to start off with Guide to LaTeX by Helmut Lopka and Patrick Daly. It's got the same publisher (Addison-Wesley) and is a book about using LaTeX. The LaTeX companion (from what I gathered by perusing it in the bookstore earlier tonight) is more about packages that provide additional functionality to the base system.
I've only had a chance to scan through bits and pieces since I brought it home, but Guide to LaTeX seems to be an excellent book that is a great starting place for learning about LaTeX.
If I had more disposable income I would have bought both of them. I've already seen places in the Guide that refer to the Companion.
I don't think that I would put "Berkeley DB" and "easy data storage" in the same sentence. Real-world use of bdb requires non-trivial application design...you're not going to be able to slap something together like you can with Access. It works great, but it can take some effort.
I've only used the C API, and maybe some of the available wrappers (perl, python) make it more developer friendly. I wouldn't recommend bdb as an answer to this guy's scenario.
Mind you, if it really worked, I'd use it. I spend a lot of time tweaking, configuring, testing, and researching to get things working just right on my computers. When I sit down to play a game, I want to just put the damn CD in and play it. I don't want to have to do anything else. I want to play a game -- not engage in some research and config fest; I'm trying to play a game because I'm tired of "making things work".
When I do play it, I don't want it to work like some mutant offspring -- "Well, the cinematics don't work" -- "It only works well in 800x600" -- "Go to your X config file and type ${whatever}" -- "The sound? Ahh, it's not *that* important, is it?"
End result? I subscribe to Cedega, and have a big old stack of games that I don't play.
I am reminded by a quote from Paul Graham.
"But VCs are mistaken to look for the next Microsoft, because no startup can be the next Microsoft unless some other company is prepared to bend over at just the right moment and be the next IBM."
I've been using graphical UIs since the early 80s, and I still get mildly frustrated a few times a day with "must click this, that and the-other" instead of "typetypetype -- done". At least clicking is "digital". I can't imagine trying to use a 3D filesystem UI where my mouse's scroll wheel (or movement or any other "analog" input) is a significant input device. "/var.../var/log/apache/...shit.../var/log...OK, easy now, easy.../var/log/apache/2002/10/httpd.log...FUCK FUCK SHIT!! *SLAM* *SLAM* *SLAM*...nfs:/moogie/home/cvs...AAAARGH!!!"
I will gladly admit that I'm very command-line-centric. The first time I tried to use a Macintosh, I whined "But, but...where are all the files!?". My first home PC was a Kim-1. (Hah! When I was a kid, we didn't *have* keyboards! My idea of intuituve input was wiring switches to the serial port)!
Someone has received just a few too many cease and desist letters.
Ahhh. But what do you define as a system? I, personally, am interested in almost all complex systems -- be it beer-making, groups of people, a person, cooking, computers, agriculture, languages, what-have-you.
I define a complex system as a system with behaviors that I will never be able to 100% accurately predict for any random period of time.
The more control I can exert over J. Random Complex System, the more likely I am to actually tinker with it. Global economics? I can't exert too much control so I'm only interested in a theoretical manner. The beer in the garage; the fields out on the acreage; the computers in the computer room? I can exert a lot of control on those systems so I'm interested in a practical manner.
I'm interested and tinker with a system if I can make inputs into a complex system and claim the output as my own. Systems where I can say, "It worked because I made it," or where I have to say, "I fucked it all up."
I'm not sure if I would look at "creating and releasing NIS into the wild" as the act of a totally benevolent corp. ;)
It must have come fast on the heels of the following commit message that so enthralled me:
From the changelog:
"1/10/2003: Replaced our TCP/IP stack with one licensed from Microsoft. Work continues on porting over the Linux virtual memory management system. "
No wonder I missed it.
*grumbles at the trolls -- even the funny ones*
Sheesh. A paltry 3 billion years. Those pathetic simulations. =]
People that don't experience anxiety or fear in combat are either (a) too stupid to know better...which means they'll probably be dead soon or (b) are probably clinically insane.
Now, don't get me wrong -- combat is the most intense adrenaline rush that you can possibly experience. It's the real-world version of "fight or flight" over and over and over again. Continuously...for hours and days and weeks. And for some people, that adrenaline rush is enjoyable and addictive. But, as an avowed adrenaline junkie, combat is an adrenaline overdose.
There is a military axiom "Train as you intend to fight". Military training is intended to ingrain patterns of behavior that soldiers will fall back on when their capacity for rational thought has been eroded by the stresses of combat. (I will not digress into a discussion of how that training is mostly centered around "how to kill and survive" rather that "how to comply with the Geneva convention and related accords" -- though that might explain why every army has a history of military atrocities).
A computer assistant -- to be effective in combat -- would have to "learn" its user in situations equivalent to combat. If it didn't, it wouldn't react the same in actual combat. And, I assure you, no training simulation comes close to approaching combat. The combination of "If I make a mistake I will die" and "I am actually killing a person" and "I could randomly die no matter what the fuck I do" is not something that can be realistically duplicated outside of the experience. I would also like to point out that soldiers do *not* respond well to "things outside of normal training parameters". The general inclination is to "shoot it" or "evade detection while trying to decide if I should shoot it".
So what we have is (a) a computer program that has "learned" its user in situations less than combat (b) a soldier that has "learned" preset responses to external stimuli -- including the computer. If you throw the soldier+computer into combat, the computer doesn't know how to react anymore -- not correctly anyway -- because it's never "learned" about combat. The soldier will tend to listen to the computer because that's what he's been trained to do...but the computer is going to be offering suboptimal advice/help/whatever...because it doesn't "know" about combat.
There's definitely roles for computers on the battlefield. Trying to provide stress-relief is probably not one of them. Certainly it's an interesting idea. And one day, it might be plausible. But, now, it's not too feasible. I think that computing will have to evolve quite a bit more before battlefield assistants that monitor emotional states and provide feedback based on that emotional state are truly a force multiplier.