the simple fact that something that works doesn't need to be replaced... software doesn't wear out, nor obsoleted given incremental upgrades.
Software doesn't wear out by itself, but the use context around it "wears in"... makes the software less useful. For instance, when the whole world switches to IPv6, I don't want to be stuck on an IPv4-only OS.
Also, the release of new exploits requires new patching. If Microsoft adds misfeatures to the security patches that makes XP less and less worthwhile to use, you might find yourself it a tough spot: use an insecure OS, or use an productivity-decreasing OS, or upgrade to Vista.
The upgrade cycle doesn't happened in a huge wave. It is more of a constant flow.
Isn't uptake of the latest version more like a Poisson-distribution?
That is, it takes a while to rise in speed until it peaks, then it slowly drops off after having converted most, with a long tail of the remaining ones upgrading very slowly.
Or maybe it's exponentially decreasing: "everybody" rushes to get the latest version on day 0---they want the latest (-n)days---and each class of less-early adopters is $FRACTION as big as the previous.
Could it be that the graphics card can't handle Vista?
Big business is also having trouble with Vista, with even firms like Intel noting XP would remain the dominant OS within the company for the foreseeable future.
It proves you know how to learn, can follow directions, and have some discipline to pursue a long-term goal.
Being autodidact proves you can't learn?;)
Being autodidact proves you know how to learn, how to come up with your own directions, and have enough interest in what you're doing to not need discipline when you want to pursue long-term goals. Isn't that more useful?
When met with some new challenge on the job, who will win? The autodidact who is used to seeking out knowledge on his own, or the diploma-having guy who is used to being taught everything?
I'm currently planning on getting a development job I'm currently somewhat underqualified for (I lack knowledge with specific technologies). Fortunately, I can get that knowledge in my spare time. The job doesn't require a degree, but I'm getting one anyways, because although it doesn't pay well, school is fun:)
I can't speak to whether my previous employer hired me based on my grades or my open source track record, but I can speak to what my projected future employer will hire me on.
YMMV. Horses for courses. Different strokes for different folks.
I disagree, when the law is inadequate, it it time to change the law.
I agree with you. But your disagreement isn't with what the parent said.
Of course laws should be changed as the context of their application changes around them.
What shouldn't happen is judges twisting existing laws to punish people who did something morally wrong but legal. That's what the parent says.
If what Drew did was illegal under some reasonable interpretation of the law, the use the law she violated to punish her. Harassment law seems like a good place to start. If what Drew did was not illegal under some reasonable representation of the law, then twisting existing law in order to convict her means that rule of law goes out the window.
If judges can stretch the letter of the law to cover what you did, then in practice you have ex post facto law: you can be punished for something you did that wasn't illegal when you did it.
The price of the rule of law is of course that you have to let some assholes go because what they didn't was only reprehensible, not illegal.
It isn't anything special in particular. What _is_ special is doing it without receiving the instruction first.
I think reinventing the wheel is a good thing in some situations; one is when you reinvent the wheel to study it but not use it. Another is when you reinvent the wheel without having been shown what a wheel is and how it works.
It's a sign that you can think for yourself and come up with good solutions to the problems you want to solve independent of instruction.
only one out of an infinite number of possibilities involves a deity.
I don't make that assumption. Take your infinite set of possibilities. Then "d" is the subset where there is a god (god \equiv designer), and "!d" is the complement.
that existence of evidence has a direct involvement with the existence of a deity.
I'm not claiming it does. What I'm saying that if there's no evidence against, then there's also no evidence for.
Your math is incorrect somewhere
No. Or at least, show me where and why.
Your conclusion would disprove any possible outcome of any possible hypothesis about anything unless there was evidence that everything we know is false.
Please show me how that follows.
My theorem doesn't say which statements about the world are true or false. It only states that there can't be evidence for unless there can also be evidence against.
your examples would seem to be consistent with what I said...
But the examples are completely interchangeable.
Instead of a blue sky of many colors, make it monochrome. It now has an entropy of 0.
Make the face on mars have many different shades of red. Maybe you can't crank the entropy up to 16 or so, but I imagine you can use 256 different shades of red (that's only 6 to 7 different values per color component), for an entropy of up to 8.
And I know that Shannon entropy isn't thermodynamic entropy. If we're talking about thermodynamic entropy, it isn't math but physics that tells us what we want to know.
And then I'm blank; that's not part of my physics training, and if I do any physics I'd rather do something applicable in my life: electronics, sound or Newtonian mechanics---in my tetris clone, blocks should fall at almost 10 m/s/s;)
RMS and FSF are really pushing it too far. Their agenda subconsciously is to prevent people from making money from their work.
Every time I've heard RMS say anything, he's explicit about freedom 0 (the freedom to share) also covering sharing in return for money (i.e. selling). He's been selling emacs support: if you paid him $200 per hour and he'd hack up whatever you wanted. The FSF has been paying people to write pieces of GNU (in RMS terminology). ISTR RMS saying he STR grep was written in this way. You can buy GNU/FSF-branded merchandise and a signed copy of "Free Software, Free Society" on either gnu.org or fsf.org. All the articles in FSFS are also on gnu.org...
"Oh noes RMS hates monies". That's flat out wrong, and the records shows it very clearly to everyone who looks at it.
In other words, a choice between two open-source drivers is more freedom than the choice between two proprietary drivers if, and only if you can make the open-source goods fit your needs.
[Stallman's philosophy that "A free system must not assist users in obtaining nonfree information" is ridiculous. How does this promote freedom?]
Freedoms conflict; you have to decide which you think are the most important freedoms. Stallman's philosophy promotes the freedoms he thinks are the most important. Your philosophy promotes the freedoms you think are important. That doesn't make any one of them wrong or ridiculous; it simply means that the two of you rank freedoms by importance differently.
But at the end of the day, it's a value judgment. Those are inherently outside the realm of right and wrong.
Consider this: if your country outlaws slavery, then your personal liberty is protected. But it costs you a different freedom: you're not free to enter into a one-sided contract that in practice would make you a slave.
Your freedom to kill people has been severely limited. In return, other people can't kill you. You're not allowed to drive in any speed you want on public roads; in return, you get some traffic safety.
For me the issue is not, do I get the source code or not? [...] But, if I do not have the right to hack it
Back in the day, Battle for Wesnoth didn't do anything with horizontal scroll events. It had a scheme for moving your viewport around in 2D using only one scroll wheel, which sucked and was hard to figure out. I fixed that; when it sees a horizontal scroll, it switches to the intuitive one-map-axis-per-wheel-axis. I would've hated to fix that without source, and I would've hated keeping using a broken scroll model.
In Nexuiz, at the time, there was no way to handicap yourself (to make the game fun against much weaker opponents). I wrote three lines of Quake C; now you can. How would I do that without source?
If you use sshfs, you might have noticed that it clears all port forwarding; if you've read the manual, you might also know that there's no option to disable it. I actually want sshfs to do port and X forwarding; what do I do? Grab the source code, grep for ClearAllForwardings, comment out four lines, off I go.
In most cases, being able to hack stuff requires source code. In all cases, it makes it a hell of a lot easier; often, so much that it goes from "infeasible" to "very easy".
not having the freedom to run it on your hardware because it's completely useless?
Wrong. I have the freedom to run gentoo on my rusty old P1 100MHz box with a 2 gig disk. It isn't going to work; the portage tree itself (~= package lists and install scripts) takes up five gigs. Compiling GNU Hello would take too long even if I do it --without-mail-reader.
What I lack is the ability to make it work and make it useful, not the freedom to do it.
I can buy a system that will work with what I have, and then have complete freedom. Buying software liberation seems to be harder than buying the right hardware for your free software.
But if you listen to the FSF definition, consider alternate definitions, viewpoints and sets of values, think for yourself about the definitions, viewpoints and values, and then make up your own mind based on what you have heard and thought---and not just by picking one, unless that's the conclusion you came to half-way independently---you have not only kept your freedom, but actively exercised it.
Find all the copy constructors he wrote and rename the argument "dolly". Find all the operator < (const T&); methods he wrote and rename the argument "dicksize".
Why are so many of you living with such FEAR of your management?
Maybe because during their careers the management has been full of archetypal suits who didn't get the joke and had a mentality that you must be productive all the time or be fired?
I'm just speculating, but I think that those people are out there somewhere.
No. Making nightly copies in date-tagged directories with symlinks for named versions is an adequate version control system. Tools like (cvs|svn|git|hg|bzr|mtn|darcs) are good version control systems.
If you write a C++ program and compile it down to one architecture, how many users do you have?
Almost all of them. All the world's an x86. Or it's an ARM. Oh noes I can't play World of Warcraft on my shiny new router:(
If you want to develop for the ARM, run stuff in an emulator and take the speed hit. Buy the hardware. Run nightly builds to check for architecture compatibility. Run daily unit tests to check for something resembling real portability. [from a cron job, of course].
If you write a browser, OS, architecture neutral application and make it available to everyone online, your user base skyrockets dramatically.
Got an example of those? Got an example that works in links? Got an example that works in links that can't be considered "ported"?
Writing architecture-independent code is fairly easy: don't stuff pointers into your ints and vice versa, let your compiler translate times-two-to-the-n into 1-shift-n for you, and don't assume pointers are always even. The whole API (libc or std::*) is portable.
Most (and I emphasis that I don't claim all) "real" problems have a natural solution that's architecture independent. OS independence is not quite as trivial; SDL gives you a portable API for sockets and threads. I recall using gaim on windows--it looked fine. I can't speak about using GTK on OS X, but for Windows+Linux you have a portable GUI library. It's not hard to write your own mini-library to handle different kinds of path name rules, and a thin layer around that which gives you the path to your config file(s). Put the rest of the paths in there. And libc can automatically handle newlines for you.
Writing portable C++ is not too hard if you know what to look out for.
Writing cross-browser portable code is hard, because the "compilers" (browsers) each accept slightly (or, in some cases, vastly) different languages. The libraries are inconsistently and partially implemented. None of the browsers have "-ansi -pedantic -Wall -Wextra -Werror".
Most of the portability issues of C++ can be tested in an easy-to-automate way. I don't know how to implement #'looks-rightp, so I can't very well test that.
You can probably set up and test server that accepts XmlHttpRequests and do some client-server inversion; have the client do a "/begin_test" request, have the server query the client in its response and have the client respond in its next query. Meh, that's ugly.
Besides: I see shitty-looking (or even worse: shitty-acting) web applications much more often than I see shitty-looking or -acting Linux applications. Look at the Debian project; while portability numbers weren't perfect last I checked, they got an ungodly large number of packages for a large number of architectures. KDE wants to port its applications to Windows.
In the real world, C++ portability is not a problem. In the real world, web app portability is harder, and as you can predict from that not done as well on average.
If we take the radio waves or terrain features we observe to be the outcome of random variables, then we're on soft ground here.
Entropy, in the mathematical sense (and specifically Shannon entropy), is a property of a random variable--that is, of a distribution function. Specifically it's defined as - sum_{all i} of (p_i * log p_i). If we've sampled the variable and observed event i, what does that tell us about p_i? About p_j for j != i?
We might take the n-pixel picture to be a random variable instead of a sample, by defining colors as outcomes and each color having probability p/n where p is the number of pixels of color p. But is that useful?
I can draw the face on mars with the mandelbrot set tattooed to its forehead in black and white (block things up and use faked shades of grey); that's going to have low entropy (at most log_2 of 2, i.e. 1). I could draw the blue sky with each pixel having a slightly different shade of blue. For a 400 by 300 picture, that'd require using 50 different values of each of R, G and B; stick them in intervals starting at 100, 150 and 200 for some reasonable values. That's going to have entropy at most log_2(400*300) = 16.87.
Admitted, my examples are retarded and chosen to screw with the suggested metric. How well would it work in practice? I don't know. How do we estimate the false negative rate for finding aliens?;)
If probability theory is a reasonable model of the existence of the ID and the outcome of observations, you can't have evidence for unless you allow evidence against.
Slashdot Mirror
the simple fact that something that works doesn't need to be replaced ... software doesn't wear out, nor obsoleted given incremental upgrades.
Software doesn't wear out by itself, but the use context around it "wears in"... makes the software less useful. For instance, when the whole world switches to IPv6, I don't want to be stuck on an IPv4-only OS.
Also, the release of new exploits requires new patching. If Microsoft adds misfeatures to the security patches that makes XP less and less worthwhile to use, you might find yourself it a tough spot: use an insecure OS, or use an productivity-decreasing OS, or upgrade to Vista.
The upgrade cycle doesn't happened in a huge wave. It is more of a constant flow.
Isn't uptake of the latest version more like a Poisson-distribution?
That is, it takes a while to rise in speed until it peaks, then it slowly drops off after having converted most, with a long tail of the remaining ones upgrading very slowly.
Or maybe it's exponentially decreasing: "everybody" rushes to get the latest version on day 0---they want the latest (-n)days---and each class of less-early adopters is $FRACTION as big as the previous.
Why would 70% of eeePC sales be XP models?
Could it be that the graphics card can't handle Vista?
Big business is also having trouble with Vista, with even firms like Intel noting XP would remain the dominant OS within the company for the foreseeable future.
But I thought Intel was Vista Capable?
It proves you know how to learn, can follow directions, and have some discipline to pursue a long-term goal.
Being autodidact proves you can't learn? ;)
Being autodidact proves you know how to learn, how to come up with your own directions, and have enough interest in what you're doing to not need discipline when you want to pursue long-term goals. Isn't that more useful?
When met with some new challenge on the job, who will win? The autodidact who is used to seeking out knowledge on his own, or the diploma-having guy who is used to being taught everything?
I'm currently planning on getting a development job I'm currently somewhat underqualified for (I lack knowledge with specific technologies). Fortunately, I can get that knowledge in my spare time. The job doesn't require a degree, but I'm getting one anyways, because although it doesn't pay well, school is fun :)
I can't speak to whether my previous employer hired me based on my grades or my open source track record, but I can speak to what my projected future employer will hire me on.
YMMV. Horses for courses. Different strokes for different folks.
I disagree, when the law is inadequate, it it time to change the law.
I agree with you. But your disagreement isn't with what the parent said.
Of course laws should be changed as the context of their application changes around them.
What shouldn't happen is judges twisting existing laws to punish people who did something morally wrong but legal. That's what the parent says.
If what Drew did was illegal under some reasonable interpretation of the law, the use the law she violated to punish her. Harassment law seems like a good place to start. If what Drew did was not illegal under some reasonable representation of the law, then twisting existing law in order to convict her means that rule of law goes out the window.
If judges can stretch the letter of the law to cover what you did, then in practice you have ex post facto law: you can be punished for something you did that wasn't illegal when you did it.
The price of the rule of law is of course that you have to let some assholes go because what they didn't was only reprehensible, not illegal.
ed(1) is a decent editor which works fine in EMACS, but it could use a better operating system.
It isn't anything special in particular. What _is_ special is doing it without receiving the instruction first.
I think reinventing the wheel is a good thing in some situations; one is when you reinvent the wheel to study it but not use it. Another is when you reinvent the wheel without having been shown what a wheel is and how it works.
It's a sign that you can think for yourself and come up with good solutions to the problems you want to solve independent of instruction.
be able to run Aero. Running Crysis was just a way of demonstrating the capability.
I think running Aero at would be a better way to demonstrate that capability.
At one frame per meeting, you're at least better off than people who play Quake over email ;)
I suggest doing everything possible to make reiser4 really really fast and have lots of bells and whistles. Then we should rename it to RicerFS.
only one out of an infinite number of possibilities involves a deity.
I don't make that assumption. Take your infinite set of possibilities. Then "d" is the subset where there is a god (god \equiv designer), and "!d" is the complement.
that existence of evidence has a direct involvement with the existence of a deity.
I'm not claiming it does. What I'm saying that if there's no evidence against, then there's also no evidence for.
Your math is incorrect somewhere
No. Or at least, show me where and why.
Your conclusion would disprove any possible outcome of any possible hypothesis about anything unless there was evidence that everything we know is false.
Please show me how that follows.
My theorem doesn't say which statements about the world are true or false. It only states that there can't be evidence for unless there can also be evidence against.
your examples would seem to be consistent with what I said...
But the examples are completely interchangeable.
Instead of a blue sky of many colors, make it monochrome. It now has an entropy of 0.
Make the face on mars have many different shades of red. Maybe you can't crank the entropy up to 16 or so, but I imagine you can use 256 different shades of red (that's only 6 to 7 different values per color component), for an entropy of up to 8.
And I know that Shannon entropy isn't thermodynamic entropy. If we're talking about thermodynamic entropy, it isn't math but physics that tells us what we want to know.
And then I'm blank; that's not part of my physics training, and if I do any physics I'd rather do something applicable in my life: electronics, sound or Newtonian mechanics---in my tetris clone, blocks should fall at almost 10 m/s/s ;)
And then the liberal arts major stands up and everybody laughs at him before he can say anything.
At least we all have some common ground ;)
RMS and FSF are really pushing it too far. Their agenda subconsciously is to prevent people from making money from their work.
Every time I've heard RMS say anything, he's explicit about freedom 0 (the freedom to share) also covering sharing in return for money (i.e. selling). He's been selling emacs support: if you paid him $200 per hour and he'd hack up whatever you wanted. The FSF has been paying people to write pieces of GNU (in RMS terminology). ISTR RMS saying he STR grep was written in this way. You can buy GNU/FSF-branded merchandise and a signed copy of "Free Software, Free Society" on either gnu.org or fsf.org. All the articles in FSFS are also on gnu.org...
"Oh noes RMS hates monies". That's flat out wrong, and the records shows it very clearly to everyone who looks at it.
In other words, a choice between two open-source drivers is more freedom than the choice between two proprietary drivers if, and only if you can make the open-source goods fit your needs.
What you are free to do is different from what you are able to do. See also http://linux.slashdot.org/comments.pl?sid=1045291&cid=25924667.
Also, what does "more freedom" really mean? Isn't the important thing whether you have the freedoms you value the most? See also http://linux.slashdot.org/comments.pl?sid=1045291&cid=25924923
[Stallman's philosophy that "A free system must not assist users in obtaining nonfree information" is ridiculous. How does this promote freedom?]
Freedoms conflict; you have to decide which you think are the most important freedoms. Stallman's philosophy promotes the freedoms he thinks are the most important. Your philosophy promotes the freedoms you think are important. That doesn't make any one of them wrong or ridiculous; it simply means that the two of you rank freedoms by importance differently.
But at the end of the day, it's a value judgment. Those are inherently outside the realm of right and wrong.
Consider this: if your country outlaws slavery, then your personal liberty is protected. But it costs you a different freedom: you're not free to enter into a one-sided contract that in practice would make you a slave.
Your freedom to kill people has been severely limited. In return, other people can't kill you. You're not allowed to drive in any speed you want on public roads; in return, you get some traffic safety.
See also http://en.wikipedia.org/wiki/Liberal_paradox (it's off on a tangent, but who doesn't love a healthy dose of game theory in the morning? ;))
For me the issue is not, do I get the source code or not? [...] But, if I do not have the right to hack it
Back in the day, Battle for Wesnoth didn't do anything with horizontal scroll events. It had a scheme for moving your viewport around in 2D using only one scroll wheel, which sucked and was hard to figure out. I fixed that; when it sees a horizontal scroll, it switches to the intuitive one-map-axis-per-wheel-axis. I would've hated to fix that without source, and I would've hated keeping using a broken scroll model.
In Nexuiz, at the time, there was no way to handicap yourself (to make the game fun against much weaker opponents). I wrote three lines of Quake C; now you can. How would I do that without source?
If you use sshfs, you might have noticed that it clears all port forwarding; if you've read the manual, you might also know that there's no option to disable it. I actually want sshfs to do port and X forwarding; what do I do? Grab the source code, grep for ClearAllForwardings, comment out four lines, off I go.
In most cases, being able to hack stuff requires source code. In all cases, it makes it a hell of a lot easier; often, so much that it goes from "infeasible" to "very easy".
not having the freedom to run it on your hardware because it's completely useless?
Wrong. I have the freedom to run gentoo on my rusty old P1 100MHz box with a 2 gig disk. It isn't going to work; the portage tree itself (~= package lists and install scripts) takes up five gigs. Compiling GNU Hello would take too long even if I do it --without-mail-reader.
What I lack is the ability to make it work and make it useful, not the freedom to do it.
I can buy a system that will work with what I have, and then have complete freedom. Buying software liberation seems to be harder than buying the right hardware for your free software.
But if you listen to the FSF definition, consider alternate definitions, viewpoints and sets of values, think for yourself about the definitions, viewpoints and values, and then make up your own mind based on what you have heard and thought---and not just by picking one, unless that's the conclusion you came to half-way independently---you have not only kept your freedom, but actively exercised it.
Find all the copy constructors he wrote and rename the argument "dolly". Find all the operator < (const T&); methods he wrote and rename the argument "dicksize".
Why are so many of you living with such FEAR of your management?
Maybe because during their careers the management has been full of archetypal suits who didn't get the joke and had a mentality that you must be productive all the time or be fired?
I'm just speculating, but I think that those people are out there somewhere.
if a tool gets the job done, it's a good tool
No. Making nightly copies in date-tagged directories with symlinks for named versions is an adequate version control system. Tools like (cvs|svn|git|hg|bzr|mtn|darcs) are good version control systems.
If you write a C++ program and compile it down to one architecture, how many users do you have?
Almost all of them. All the world's an x86. Or it's an ARM. Oh noes I can't play World of Warcraft on my shiny new router :(
If you want to develop for the ARM, run stuff in an emulator and take the speed hit. Buy the hardware. Run nightly builds to check for architecture compatibility. Run daily unit tests to check for something resembling real portability. [from a cron job, of course].
If you write a browser, OS, architecture neutral application and make it available to everyone online, your user base skyrockets dramatically.
Got an example of those? Got an example that works in links? Got an example that works in links that can't be considered "ported"?
Writing architecture-independent code is fairly easy: don't stuff pointers into your ints and vice versa, let your compiler translate times-two-to-the-n into 1-shift-n for you, and don't assume pointers are always even. The whole API (libc or std::*) is portable.
Most (and I emphasis that I don't claim all) "real" problems have a natural solution that's architecture independent. OS independence is not quite as trivial; SDL gives you a portable API for sockets and threads. I recall using gaim on windows--it looked fine. I can't speak about using GTK on OS X, but for Windows+Linux you have a portable GUI library. It's not hard to write your own mini-library to handle different kinds of path name rules, and a thin layer around that which gives you the path to your config file(s). Put the rest of the paths in there. And libc can automatically handle newlines for you.
Writing portable C++ is not too hard if you know what to look out for.
Writing cross-browser portable code is hard, because the "compilers" (browsers) each accept slightly (or, in some cases, vastly) different languages. The libraries are inconsistently and partially implemented. None of the browsers have "-ansi -pedantic -Wall -Wextra -Werror".
Most of the portability issues of C++ can be tested in an easy-to-automate way. I don't know how to implement #'looks-rightp, so I can't very well test that.
You can probably set up and test server that accepts XmlHttpRequests and do some client-server inversion; have the client do a "/begin_test" request, have the server query the client in its response and have the client respond in its next query. Meh, that's ugly.
Besides: I see shitty-looking (or even worse: shitty-acting) web applications much more often than I see shitty-looking or -acting Linux applications. Look at the Debian project; while portability numbers weren't perfect last I checked, they got an ungodly large number of packages for a large number of architectures. KDE wants to port its applications to Windows.
In the real world, C++ portability is not a problem. In the real world, web app portability is harder, and as you can predict from that not done as well on average.
I don't have any of those, go fish...
The math would agree with you. [...] entropy
If we take the radio waves or terrain features we observe to be the outcome of random variables, then we're on soft ground here.
Entropy, in the mathematical sense (and specifically Shannon entropy), is a property of a random variable--that is, of a distribution function. Specifically it's defined as - sum_{all i} of (p_i * log p_i). If we've sampled the variable and observed event i, what does that tell us about p_i? About p_j for j != i?
We might take the n-pixel picture to be a random variable instead of a sample, by defining colors as outcomes and each color having probability p/n where p is the number of pixels of color p. But is that useful?
I can draw the face on mars with the mandelbrot set tattooed to its forehead in black and white (block things up and use faked shades of grey); that's going to have low entropy (at most log_2 of 2, i.e. 1). I could draw the blue sky with each pixel having a slightly different shade of blue. For a 400 by 300 picture, that'd require using 50 different values of each of R, G and B; stick them in intervals starting at 100, 150 and 200 for some reasonable values. That's going to have entropy at most log_2(400*300) = 16.87.
Admitted, my examples are retarded and chosen to screw with the suggested metric. How well would it work in practice? I don't know. How do we estimate the false negative rate for finding aliens? ;)
If probability theory is a reasonable model of the existence of the ID and the outcome of observations, you can't have evidence for unless you allow evidence against.
See my proof at http://science.slashdot.org/comments.pl?sid=1045125&cid=25918745