Now that I think about it, I must say, please don't quote me out of context. I was restating his point on purpose. And then made my own in the following sentence:
This is not a bad thing, because Linux is flexible enough to support nearly any kind of computation environment, right now. Including the weird experimental ones he (and I) like. And the run of the mill workhorse desktop environments most people need.
I think this is exactly the point of the article- 30 years of "best practices" and the best open source can come up with is a Unix clone (cloning a 30 year old OS model)?
The phrasing here is nearly flamebait. I don't particularly care, but it shows your bias. Linux is a solid kernel that performs its job well.
Operating systems are just scaffolding. Once they've reached a certain level of flexibility, there's no point in changing their external programmatic interfaces, since at that point they can ALREADY RUN ANY CRAZY EXPERIMENTAL COMPUTATION ENVIRONMENT YOU WANT. Squeak? KDE? Dr. Scheme? Gnome? Haskell? Cocoa? They're all just ways to tell your computer what to do. Their internals might be tied to the Unix model, but their external interface (the one you see) certainly does not have to be. Some of them were borne from academic research, including Squeak, Lisp, and Haskell.
OS X is a Unix, and it's innovative. Why? Because NeXT came up with an interesting programming environment and sold it to Apple. OS X would have been just as innovative using the NT kernel. It isn't the kernel's job to make a computer awesome.
Linux is essentially irrelevant to the topic of open source innovation. Indeed, it is a red herring.
Sure, there are *popular* examples, such as Apache. But popularity doesn't mean innovative. Apache was simply one of the first web servers, which caused it to get hammered on until it was useful. But there's nothing in Apache that makes you stand back and say, "Wow! That's absolutely brilliant thinking!"
If you're cynical enough, you could say the same thing about any software. On the other hand, Apache was innovative. And the Apache Foundation continues to found and fund new projects, including SpamAssassin -- the first Bayesian spam filter.
In any case, Haskell is open source. So is Erlang.
While I'm sympathetic to Jaron's point, I think he's missing a big one. Linux represents about 30 years of knowledge of best practices in software engineering. This is not a bad thing, because Linux is flexible enough to support nearly any kind of computation environment, right now. Including the weird experimental ones he (and I) like. And the run of the mill workhorse desktop environments most people need.
2. There are many more ways for cables to be tangled than to be untangled, so statistically, tangling is overwhelmingly likely. It's like entropy that way: There are many more ways for particles to move in different directions than there are ways for particles to move in the same direction, so it takes special effort or special circumstances to get them all to line up.
You need to make the notion of counting ways to be tangled and untangled more precise. In any case, the problem with real cables is that most cable runs have a half turn in them. But where the turn happens varies. Moreover, the turn introduces distortion in the cable at the turn since it isn't under tension. Heating and cooling, and Type I and II Reidemeister moves caused by the distortion moving do the rest.
But note that these kinds of knots are trivial to untangle if you keep the cables connected, and much harder if you don't, since Type I and II Reidemeister moves can't produce knots, just tangles.
I agree, and don't think you should have been modded flamebait, though I can see why some people might think it is. Special cases are anathema to good interface design. Things that are by all appearances of the same kind should be treated the same by the interface. Each special case introduces a new context the user has to learn. Special cases that change with every release are even worse.
I don't have a problem with limited configurability, if the defaults are sane. Special cases aren't. KDE has some problems with special cases too, but they're easy enough to configure away if they become a nuisance. I've only run into one special case problem in Leopard, and it was hard to fix -- but honestly no harder to fix than a Gnome usability issue would be to a noob. I'm referring to how the Dock handles folders. In Tiger and before, clicking a folder would open a Finder window for the folder. Now "Stacks" open by default, a layer of indirection I don't want.
You're advocating a false economy. Sure, capabilities are potentially complex and the implementation is hard to get right. Still, you only have to do it right one time and everything else benefits. In today's situation, we have things like "ping" running setuid root so that it can create raw sockets and Apache launching as root so that it can bind port 80 before dropping privileges.
Yup. If Debian came out with an apt package that implemented ACLs that just worked with apt, dozens of people around the world would use it. Ubuntu would probably pick it up and turn it on by default (since less flexibility is needed on the desktop), leading to hundreds of people benefitting.
If you want the punch line, 1/n converges to 0 and 01/n
Limits are unique in metric spaces. So if it converges to 0, and it converges to (0*1)/n, you know they're the same: the rational function 0. The set of rational functions {1/n} has a greatest lower bound -- the rational function 0.
I still don't see where you're going with this. The structure you described is either isomorphic to the real line, because of the real number axiom's categoricity, or the structure is not isomorphic because it's not totally ordered. I will admit that conventional language here is ambiguous. I meant a total order. If this is what your objection referred to, fine, I will agree that a merely partially ordered complete field is not isomorphic to the reals.
Since this started your objection, note that:
A complete metric is a metric in which every Cauchy sequence is convergent. A topological space with a complete metric is called a complete metric space.http://mathworld.wolfram.com/CompleteMetric.html
I have a very nice halogen desk lamp next to me. If I was unable to get bulbs for it in the future, I would have to turn the heat up in my room (at least -- as a matter of brute fact I would have to turn it up in my entire home since I don't have localized heating). There's no such thing as waste heat in the winter. And as a bonus, this light gives me a nice rich spectrum to work with.
Better yet, an explicit proof. Assume that each bounded set has a least upper bound.
First: the range of a Cauchy sequence is bounded.
Pf/ Let {x_n} be a Cauchy sequence and epsilon > 0. Then there is an N such than m and m => N implies that d(x_m, x_n) epsilon. Since N is finite, there is a maximum pairwise distance between points whose indices are less than or equal to N, denoted max_{i,j=N} {d(x_i, x_j)}. Thus the maximum distance is max_{i,j=N} {d(x_i, x_j)} + epsilon.
Second: dually to the condition that each bounded set has a least upper bound, each bounded set has a greatest lower bound.
Third: Since {x_n} is bounded, d(x_i, x_j) is bounded as i and j vary. Thus the set has a greatest lower bound. Which is clearly 0 since the sequence is Cauchy.
To prove that they are equivalent, note that there is a bijective correspondence between elements of convergent sequences and least upper bounds on finite sets. Use the face that the sequence is Cauchy.
Zermelo-Frankel Set Theory is an axiomatization of set theory. That is to say, it is a list of axioms describing properties of any structure that is meant to be a collection of sets. There are alternative structures and alternative axiomatizations to generate those structures. (FYI, a consequence of Godel's Incompleteness Theorem is that there are infinitely distinct (in a non-trivial sense) axiomatizations of the natural numbers.)
Since you've studied Diff Eqs, I'll give you a little example of why axioms of this kind are needed. You were studying differentiable functions. Many of their properties are due to the completeness of the real numbers. Many of their properties are due the real numbers being ordered. Some are due to the fact that the real numbers form a field. And while tools like linear algebra might be necessary to study differential equations, all the properties of differentiable functions are caused by at least one of these three (and the definition of a differentiable function).
It turns out the real numbers can be characterized as the complete ordered field. To axiomatize the real numbers -- to write sentences from which all the others follow -- we just have to group together the completeness axiom (Every Cauchy sequence converges in the set), the field axioms, and the order axioms. If, for example, you drop the completeness axiom, you would also be writing about things like the rational numbers since they're an ordered field that happens to not be complete.
Axioms aren't about truth. They have a specific meaning in logic, and more importantly act as a sign post to the audience saying: this is what I'm going to talk about, and how I'm going to talk about it. Of course, in practice, mathematicians don't explicitly state these axioms unless they are the subject of the paper. But they are implicitly "contained" in the jargon.
Upvar/Uplevel are fantastic when you understand what they're used for. Uplevel enables tcl to be completely extensible - you can write new tcl language elements within tcl. For example, one can write a brand new transaction command which wraps its contents in a db transaction open/close pair, and catches errors to abort transactions. I really miss that when I have to code in other languages.
Sounds a lot like Perl's Ties. Are you familiar with them? If so, how do they compare?
Re:It's possible - using a very thick concrete wal
on
Specs For the New KITT
·
· Score: 1
I strongly suggest you send a polite email to your school's IT staff explaining your academic needs. If that doesn't work, ask your professors to speak with the IT staff on your behalf. Don't let talk of 'policy' get in the way. Talk to policy makers on campus and get them to talk to IT on your behalf. Don't be rude. This isn't a crusade. You (presumably) have legitimate academic needs and a legitimate academic use for the resources IT would grant. That's their job.
I find it depressing when students let shit like this happen to them.
Yes. Luckily, aftermarket RAM is much cheaper. I bought 4 GB for $109 at Newegg, for my brand new iMac C2D. (Despite being a desktop, this is still relevant. iMacs use the same RAM as the MBPs)
Um.. No. The right strings have to be pressed when you strum.. The guitar, for the most part, doesn't care when you pushed 'em down.
Um.. Yes. The right strings have to be pressed down when you run the plectrum over them. If you're playing arpeggios, getting the timing right is a lot harder than playing a plain old chord. This is true of most forms of guitar playing, by the way.
Slashdot Mirror
1700m is over a mile.
I think this is exactly the point of the article- 30 years of "best practices" and the best open source can come up with is a Unix clone (cloning a 30 year old OS model)?
The phrasing here is nearly flamebait. I don't particularly care, but it shows your bias. Linux is a solid kernel that performs its job well.
Operating systems are just scaffolding. Once they've reached a certain level of flexibility, there's no point in changing their external programmatic interfaces, since at that point they can ALREADY RUN ANY CRAZY EXPERIMENTAL COMPUTATION ENVIRONMENT YOU WANT. Squeak? KDE? Dr. Scheme? Gnome? Haskell? Cocoa? They're all just ways to tell your computer what to do. Their internals might be tied to the Unix model, but their external interface (the one you see) certainly does not have to be. Some of them were borne from academic research, including Squeak, Lisp, and Haskell.
OS X is a Unix, and it's innovative. Why? Because NeXT came up with an interesting programming environment and sold it to Apple. OS X would have been just as innovative using the NT kernel. It isn't the kernel's job to make a computer awesome.
Linux is essentially irrelevant to the topic of open source innovation. Indeed, it is a red herring.
Sure, there are *popular* examples, such as Apache. But popularity doesn't mean innovative. Apache was simply one of the first web servers, which caused it to get hammered on until it was useful. But there's nothing in Apache that makes you stand back and say, "Wow! That's absolutely brilliant thinking!"
If you're cynical enough, you could say the same thing about any software. On the other hand, Apache was innovative. And the Apache Foundation continues to found and fund new projects, including SpamAssassin -- the first Bayesian spam filter.
In any case, Haskell is open source. So is Erlang.
While I'm sympathetic to Jaron's point, I think he's missing a big one. Linux represents about 30 years of knowledge of best practices in software engineering. This is not a bad thing, because Linux is flexible enough to support nearly any kind of computation environment, right now. Including the weird experimental ones he (and I) like. And the run of the mill workhorse desktop environments most people need.
2. There are many more ways for cables to be tangled than to be untangled, so statistically, tangling is overwhelmingly likely. It's like entropy that way: There are many more ways for particles to move in different directions than there are ways for particles to move in the same direction, so it takes special effort or special circumstances to get them all to line up.
You need to make the notion of counting ways to be tangled and untangled more precise. In any case, the problem with real cables is that most cable runs have a half turn in them. But where the turn happens varies. Moreover, the turn introduces distortion in the cable at the turn since it isn't under tension. Heating and cooling, and Type I and II Reidemeister moves caused by the distortion moving do the rest.
But note that these kinds of knots are trivial to untangle if you keep the cables connected, and much harder if you don't, since Type I and II Reidemeister moves can't produce knots, just tangles.
I agree, and don't think you should have been modded flamebait, though I can see why some people might think it is. Special cases are anathema to good interface design. Things that are by all appearances of the same kind should be treated the same by the interface. Each special case introduces a new context the user has to learn. Special cases that change with every release are even worse.
I don't have a problem with limited configurability, if the defaults are sane. Special cases aren't. KDE has some problems with special cases too, but they're easy enough to configure away if they become a nuisance. I've only run into one special case problem in Leopard, and it was hard to fix -- but honestly no harder to fix than a Gnome usability issue would be to a noob. I'm referring to how the Dock handles folders. In Tiger and before, clicking a folder would open a Finder window for the folder. Now "Stacks" open by default, a layer of indirection I don't want.
Rule 35 is that if there's no porn of it, it's your duty to make it. Jesus fucking christ noob, lurk moar.
I agree, I don't think any application should be using resources on my system without my explicit consent.
You gave it when you explicitly agreed to the EULA.
Hot Jupiters are Jupiter (or larger) sized planets that are significantly closer to their sun than our Jupiter is.
You're advocating a false economy. Sure, capabilities are potentially complex and the implementation is hard to get right. Still, you only have to do it right one time and everything else benefits. In today's situation, we have things like "ping" running setuid root so that it can create raw sockets and Apache launching as root so that it can bind port 80 before dropping privileges.
Yup. If Debian came out with an apt package that implemented ACLs that just worked with apt, dozens of people around the world would use it. Ubuntu would probably pick it up and turn it on by default (since less flexibility is needed on the desktop), leading to hundreds of people benefitting.
Limits are unique in metric spaces. So if it converges to 0, and it converges to (0*1)/n, you know they're the same: the rational function 0. The set of rational functions {1/n} has a greatest lower bound -- the rational function 0.
I still don't see where you're going with this. The structure you described is either isomorphic to the real line, because of the real number axiom's categoricity, or the structure is not isomorphic because it's not totally ordered. I will admit that conventional language here is ambiguous. I meant a total order. If this is what your objection referred to, fine, I will agree that a merely partially ordered complete field is not isomorphic to the reals.
Since this started your objection, note that:
Whoever modded parent redundant is an idiot.
I have a very nice halogen desk lamp next to me. If I was unable to get bulbs for it in the future, I would have to turn the heat up in my room (at least -- as a matter of brute fact I would have to turn it up in my entire home since I don't have localized heating). There's no such thing as waste heat in the winter. And as a bonus, this light gives me a nice rich spectrum to work with.
I will be stocking up on halogen bulbs.
I messed up. Oh well.
Typo: d(x_m, x_n) =< epsilon
Better yet, an explicit proof. Assume that each bounded set has a least upper bound.
First: the range of a Cauchy sequence is bounded.
Pf/ Let {x_n} be a Cauchy sequence and epsilon > 0. Then there is an N such than m and m => N implies that d(x_m, x_n) epsilon. Since N is finite, there is a maximum pairwise distance between points whose indices are less than or equal to N, denoted max_{i,j=N} {d(x_i, x_j)}. Thus the maximum distance is max_{i,j=N} {d(x_i, x_j)} + epsilon.
Second: dually to the condition that each bounded set has a least upper bound, each bounded set has a greatest lower bound.
Third: Since {x_n} is bounded, d(x_i, x_j) is bounded as i and j vary. Thus the set has a greatest lower bound. Which is clearly 0 since the sequence is Cauchy.
Completeness in the real numbers is not that every Cauchy sequence converges, it is that every set bounded above has an upper bound.
You mean a supremum, or least upper bound. Having just an upper bound is trivial, since every number is finite. The "axioms" happen to be equivalent.
http://en.wikipedia.org/wiki/Complete_space
http://www.mathology.net/mathology/vis_articolo.asp?id=44&lang=ita
To prove that they are equivalent, note that there is a bijective correspondence between elements of convergent sequences and least upper bounds on finite sets. Use the face that the sequence is Cauchy.
Zermelo-Frankel Set Theory is an axiomatization of set theory. That is to say, it is a list of axioms describing properties of any structure that is meant to be a collection of sets. There are alternative structures and alternative axiomatizations to generate those structures. (FYI, a consequence of Godel's Incompleteness Theorem is that there are infinitely distinct (in a non-trivial sense) axiomatizations of the natural numbers.)
Since you've studied Diff Eqs, I'll give you a little example of why axioms of this kind are needed. You were studying differentiable functions. Many of their properties are due to the completeness of the real numbers. Many of their properties are due the real numbers being ordered. Some are due to the fact that the real numbers form a field. And while tools like linear algebra might be necessary to study differential equations, all the properties of differentiable functions are caused by at least one of these three (and the definition of a differentiable function).
It turns out the real numbers can be characterized as the complete ordered field. To axiomatize the real numbers -- to write sentences from which all the others follow -- we just have to group together the completeness axiom (Every Cauchy sequence converges in the set), the field axioms, and the order axioms. If, for example, you drop the completeness axiom, you would also be writing about things like the rational numbers since they're an ordered field that happens to not be complete.
Axioms aren't about truth. They have a specific meaning in logic, and more importantly act as a sign post to the audience saying: this is what I'm going to talk about, and how I'm going to talk about it. Of course, in practice, mathematicians don't explicitly state these axioms unless they are the subject of the paper. But they are implicitly "contained" in the jargon.
The headline was obviously written in Soviet Russia.
Upvar/Uplevel are fantastic when you understand what they're used for. Uplevel enables tcl to be completely extensible - you can write new tcl language elements within tcl. For example, one can write a brand new transaction command which wraps its contents in a db transaction open/close pair, and catches errors to abort transactions. I really miss that when I have to code in other languages.
Sounds a lot like Perl's Ties. Are you familiar with them? If so, how do they compare?
Kinetic energy = 1/2 * mass * velocity squared
http://www.google.com/search?num=100&hl=en&safe=off&client=safari&rls=en-us&q=1%2F2+*+1000+kg+*+(134.1120+m%2Fs)+squared&btnG=Search
In summation, a lot.
Guys... it's a TV show.
Uh, X11 comes with OS X. If I recall correctly, it's installed by default by the Leopard installer.
Your fanaticism is showing.
I strongly suggest you send a polite email to your school's IT staff explaining your academic needs. If that doesn't work, ask your professors to speak with the IT staff on your behalf. Don't let talk of 'policy' get in the way. Talk to policy makers on campus and get them to talk to IT on your behalf. Don't be rude. This isn't a crusade. You (presumably) have legitimate academic needs and a legitimate academic use for the resources IT would grant. That's their job.
I find it depressing when students let shit like this happen to them.
Yes. Luckily, aftermarket RAM is much cheaper. I bought 4 GB for $109 at Newegg, for my brand new iMac C2D. (Despite being a desktop, this is still relevant. iMacs use the same RAM as the MBPs)
Um.. No. The right strings have to be pressed when you strum.. The guitar, for the most part, doesn't care when you pushed 'em down.
Um.. Yes. The right strings have to be pressed down when you run the plectrum over them. If you're playing arpeggios, getting the timing right is a lot harder than playing a plain old chord. This is true of most forms of guitar playing, by the way.