1. 99% of users don't care about the operating system. The only thing that matters is the applications.
2. Stability for the desktop is way down the list of priorities for the average user. Is it nice? Yes, but it's not a big issue. [Proof: If it was, the Mac would be dead, and Windows never would have dominated.
Linux will never go anywhere on the desktop until it gets some decent applications that are at least comparably to Windows. Right now they are way, way behind. I've said it before, and I'll say it again: Name for me one significant end-user application that is clearly superior than the equivalent in Windows. There simply isn't one.
It seems to me that "meme" brings with it an interesting connotations: (1) Memes spread; (2) Memes compete with one another; (3) People study the nature of (1) and (2).
But we've always had the concept of "the spread of ideas", or "the conflict of ideas".
I suppose you could say that the study of how ideas are spread could be new, but I still don't see the need for a jargony word. I guess it helps sell books, though.:)
Every time you use the word "meme" in your post, I could substitute the word "idea" and it means exactly the same thing. The one paragraph where perhaps it is a little more awkward is:
Some of the most powerful memes can be found in things that can't be described as ideas, football (soccer to those in the US) is a meme that has a life of its own and has spread around the globe pretty much unhindered.
But even then, I could say something like "football is an idea whose time has come" and people know what I'm talking about.
But even in this definition, they use the word "idea" almost interchangeably: "Ideas can evolve in a way analogous to biological evolution. Some ideas survive better than others; ideas can mutate through, for example, misunderstandings; and two ideas can recombine to produce a new idea involving elements of each parent idea."
I'm surprised at many of you people. As much as some complain about the difference between "hacking" and "cracking", I would think you would be happy at this categorization.
To me, if they had spoofed the protocol, or had invaded CNN's servers, that would have been a "crack". Since this was more in line of a clever use of the technology, it should be charactorized as a "hack".
For once the media got it right! Rejoice!
P.S. I still reserve the future right to rail against people for pedantic use of "hack" v.s. "crack".:)
This is not really a question, but a comment on this interview. I accept that in theory this interview is supposed to be about the process of using the Internet for a political campaign, but why do I have a feeling it will turn into a political speech? Something like:
Q: What do you think the impact of the Internet will be on future campaigns?
A: Al Gore is a huge believer in the power of the Internet, and intends to be a front runner in making sure everyone has access to this powerful bridge into the 21st century so that every citizen can have the benefit of directly reaching their candidate. [insert more hot air here]
Does anyone really think that the questions will not be run through 6 different P.R. people, each spinning it a little more toward Al Gore?
If my prediction turns out to be correct, I hope that Slashdot will give other candidates a shot. In fact, equal access laws may require it.
If you look beyond that one sentence, you'll note that probably 90% of my career has been spent using Unix.
Beyond that, NT has a lot of advantages in certain situations, and Unix has advantages in other situations. I don't "advocate" either one. I engineer solutions using the appropriate technology based on the business requirements.
Why am I bothering to explain this to someone who probably isn't going to get it anyway?
umm We don't know how long the NT box was down because of the hardware failures. Since the average came down to almost 1 hour/crash, we can assume that the hardware replacement took up most of that time.
You are calling a fellow human being irrationally blind because he expressed his opinion, what does that make you? What are you making the whole NT advocacy look like with your rant?
I don't call him irrationally blind for his opinion, I call him irrationally blind for the reasoning behind his opinion. Whether the results of test happen to match real world reality is totally irrelevent.
He didn't even try to refute my point, which is that the basis of the test was flawed. He just assumes that since the test gave the result that he expected, then the test must have been perfectly executed.
And by the way, where did I say I'm an NT advocate?
Actually, I think I know why... try going into your IE settings, and turn off "use smooth scrolling". You need a fast processor and fast video card for that not to be obnoxious.
If I could extract one of the (few) valid points in the guy's article...
I'm not going to cast blame on Torvalds, since I don't think he necessarily controls anything, but if one believes in Open Source, shouldn't the Code Morphing code be released?
It would be interesting to know where RMS comes down on this, and firmware in general, such as the downloadable Microcode in Intel chips.
<rant>No, what's tired is people like you who can't apply rational thought to the article at hand. All you can do is parrot the conventional wisdom that that article must be correct if it draws the conclusion that Linux is better, while an article is crap if draws the conclusion that NT is better. If you're so blind that you can't see that this test was horribly done, then pull your head out of the advocacy sand. You just make the whole Linux advocacy look like fools.</rant>
I now return you to your regularly schedule Microsoft bashing.
What kind of crap hardware were they running? Not to necessarily defend NT, but it sounds like many of the problems could have been related to whatever hardware they picked.
Actually, believe it or not, the symbols make zero different in start-up time (at least in most Unixes, don't know about Linux). The way it usually works is that an executable is mapped into virtual memory, not real memory. Then as each part of the executable is run, the page is swapped into memory.
Note that this allows the side benefit of only one copy of an executable in memory if multiple people are running the same program.
The upshot is that if the symbols aren't used, they won't be swapped into real memory. At least on most Unixes; don't know about Linux, but I would imagine they did it the "right way".
Again, try it for yourself by doing a "strip" on the binary.
The problem with the start-up time is probably that no one has optimized how Netscape/Mozilla interacts with X11. Unfortunately, I think they don't have anyone to sit down and analyze the stream of X11 commands to optimize it. I've done it, and you can make a HUGE difference by minimizing the X11 round trips.
Exactly. Just making tables work right is an exercise in extreme frustration. I think people need to use IE5 for an extended period of time to realize just how good the layout rendering engine is. It's also amazingly fast (watch it dynamically resize tables in real time as you make the window larger and smaller. Netscape waits until your done, thinks for about 3-4 seconds, and then might or might not do it right).
Yes, the interface needs tweaking, but Gecko, the layout engine, is effectively finished....and then later... As far as people saying that Mozilla is slow, go look into the project a bit. Performance enhancements have not been rigorously done yet, and everyone is also using debug builds.
But the layout engine is where the performance problems would be! If that's already done, then where are these performance improvements supposed to come from?
Umm... maybe if it didn't take 900 megs of free space and several hours to build, you might see more "evidence" of the type you are looking for.
Several hours is not that long a time. Considering the number of complaints about the speed, someone would compile up a version.
FYI, any program compiled with debug symbols (from the largest to the smallest) gets speeded up quite a bit when those are removed (300% is probably ridiculous, but a doubling isn't completely out of the question.)
On every Unix I have used, symbols only make the binary file larger, not slower due to the way debuggers work. Think about it: What would change in the assembly language for symbols? (hint: assembly language doesn't use alphanumeric symbols).
But just to make sure Linux wasn't weird in some way, I performed the following test using an associative array library I wrote. Including the little test subroutine, it's about 800 lines of code (not that big, but should have an effect if symbols are a factor). The test subroutine basically creates an AA, adds a bunch of symbols, reads them back, destroys it, and repeats a few thousands times:
Slashdot Mirror
Two other things Linux advocates need to learn:
1. 99% of users don't care about the operating system. The only thing that matters is the applications.
2. Stability for the desktop is way down the list of priorities for the average user. Is it nice? Yes, but it's not a big issue. [Proof: If it was, the Mac would be dead, and Windows never would have dominated.
Linux will never go anywhere on the desktop until it gets some decent applications that are at least comparably to Windows. Right now they are way, way behind. I've said it before, and I'll say it again: Name for me one significant end-user application that is clearly superior than the equivalent in Windows. There simply isn't one.
--
But we've always had the concept of "the spread of ideas", or "the conflict of ideas".
I suppose you could say that the study of how ideas are spread could be new, but I still don't see the need for a jargony word. I guess it helps sell books, though. :)
--
Every time you use the word "meme" in your post, I could substitute the word "idea" and it means exactly the same thing. The one paragraph where perhaps it is a little more awkward is:
Some of the most powerful memes can be found in things that can't be described as ideas, football (soccer to those in the US) is a meme that has a life of its own and has spread around the globe pretty much unhindered.
But even then, I could say something like "football is an idea whose time has come" and people know what I'm talking about.
--
But even in this definition, they use the word "idea" almost interchangeably: "Ideas can evolve in a way analogous to biological evolution. Some ideas survive better than others; ideas can mutate through, for example, misunderstandings; and two ideas can recombine to produce a new idea involving elements of each parent idea."
--
What's wrong with the word that has meant the same thing since English began: idea?
Is it just to be extra l33t, or is there some hidden meaning that has escaped me?
--
Pranks are included in "hacks", at least according to the Jargon File. The canonical example would be the Cal Tech Rose Bowl Hack.
--
I'm surprised at many of you people. As much as some complain about the difference between "hacking" and "cracking", I would think you would be happy at this categorization.
To me, if they had spoofed the protocol, or had invaded CNN's servers, that would have been a "crack". Since this was more in line of a clever use of the technology, it should be charactorized as a "hack".
For once the media got it right! Rejoice!
P.S. I still reserve the future right to rail against people for pedantic use of "hack" v.s. "crack". :)
--
This is not really a question, but a comment on this interview. I accept that in theory this interview is supposed to be about the process of using the Internet for a political campaign, but why do I have a feeling it will turn into a political speech? Something like:
Q: What do you think the impact of the Internet will be on future campaigns?
A: Al Gore is a huge believer in the power of the Internet, and intends to be a front runner in making sure everyone has access to this powerful bridge into the 21st century so that every citizen can have the benefit of directly reaching their candidate. [insert more hot air here]
Does anyone really think that the questions will not be run through 6 different P.R. people, each spinning it a little more toward Al Gore?
If my prediction turns out to be correct, I hope that Slashdot will give other candidates a shot. In fact, equal access laws may require it.
--
I should suffer fools better, but...
If you look beyond that one sentence, you'll note that probably 90% of my career has been spent using Unix.
Beyond that, NT has a lot of advantages in certain situations, and Unix has advantages in other situations. I don't "advocate" either one. I engineer solutions using the appropriate technology based on the business requirements.
Why am I bothering to explain this to someone who probably isn't going to get it anyway?
--
umm We don't know how long the NT box was down because of the hardware failures. Since the average came down to almost 1 hour/crash, we can assume that the hardware replacement took up most of that time.
--
You are calling a fellow human being irrationally blind because he expressed his opinion, what does that make you? What are you making the whole NT advocacy look like with your rant?
I don't call him irrationally blind for his opinion, I call him irrationally blind for the reasoning behind his opinion. Whether the results of test happen to match real world reality is totally irrelevent.
He didn't even try to refute my point, which is that the basis of the test was flawed. He just assumes that since the test gave the result that he expected, then the test must have been perfectly executed.
And by the way, where did I say I'm an NT advocate?
--
Actually, I think I know why... try going into your IE settings, and turn off "use smooth scrolling". You need a fast processor and fast video card for that not to be obnoxious.
--
If I could extract one of the (few) valid points in the guy's article...
I'm not going to cast blame on Torvalds, since I don't think he necessarily controls anything, but if one believes in Open Source, shouldn't the Code Morphing code be released?
It would be interesting to know where RMS comes down on this, and firmware in general, such as the downloadable Microcode in Intel chips.
--
<rant>No, what's tired is people like you who can't apply rational thought to the article at hand. All you can do is parrot the conventional wisdom that that article must be correct if it draws the conclusion that Linux is better, while an article is crap if draws the conclusion that NT is better. If you're so blind that you can't see that this test was horribly done, then pull your head out of the advocacy sand. You just make the whole Linux advocacy look like fools.</rant>
I now return you to your regularly schedule Microsoft bashing.
--
I believe that the failures were software failures, not hardware failures. If it was just a hardware problem, they would have been solved long ago.
Correct me if I'm wrong, but the backend software for EBay was designed by Sun under contract.
--
That could be true, but the subject is the infamous EBay downtime. The point is that you can't blame Microsoft for these high-profile failures.
--
Even if they ran the same "crap" for both tests, it's still not a reasonable test. Linux just got luckier that it didn't fail.
--
What kind of crap hardware were they running? Not to necessarily defend NT, but it sounds like many of the problems could have been related to whatever hardware they picked.
--
Ironically, it's the Sun part of EBay that has constantly failed. The NT part has worked flawlessly.
--
Actually, believe it or not, the symbols make zero different in start-up time (at least in most Unixes, don't know about Linux). The way it usually works is that an executable is mapped into virtual memory, not real memory. Then as each part of the executable is run, the page is swapped into memory.
Note that this allows the side benefit of only one copy of an executable in memory if multiple people are running the same program.
The upshot is that if the symbols aren't used, they won't be swapped into real memory. At least on most Unixes; don't know about Linux, but I would imagine they did it the "right way".
Again, try it for yourself by doing a "strip" on the binary.
The problem with the start-up time is probably that no one has optimized how Netscape/Mozilla interacts with X11. Unfortunately, I think they don't have anyone to sit down and analyze the stream of X11 commands to optimize it. I've done it, and you can make a HUGE difference by minimizing the X11 round trips.
--
Exactly. Just making tables work right is an exercise in extreme frustration. I think people need to use IE5 for an extended period of time to realize just how good the layout rendering engine is. It's also amazingly fast (watch it dynamically resize tables in real time as you make the window larger and smaller. Netscape waits until your done, thinks for about 3-4 seconds, and then might or might not do it right).
--
Uh, read the thread. I didn't make the claim, he did.
--
(hint: assembly language doesn't use alphanumeric symbols).
I should say, machine language doesn't use alphanumeric symbols. Assembly language obviously does.
--
But the layout engine is where the performance problems would be! If that's already done, then where are these performance improvements supposed to come from?
--
Several hours is not that long a time. Considering the number of complaints about the speed, someone would compile up a version.
On every Unix I have used, symbols only make the binary file larger, not slower due to the way debuggers work. Think about it: What would change in the assembly language for symbols? (hint: assembly language doesn't use alphanumeric symbols).
But just to make sure Linux wasn't weird in some way, I performed the following test using an associative array library I wrote. Including the little test subroutine, it's about 800 lines of code (not that big, but should have an effect if symbols are a factor). The test subroutine basically creates an AA, adds a bunch of symbols, reads them back, destroys it, and repeats a few thousands times:
$ gcc speed.c aa.c -o speed
$ time speed
2.14user 0.01system 0:02.14elapsed 100%CPU
$ gcc -g speed.c aa.c -o speed
$ time speed
2.15user 0.00system 0:02.14elapsed 100%CPU
$ gcc -g -O speed.c aa.c -o speed
$ time speed
2.00user 0.00system 0:02.00elapsed 99%CPU
$ gcc -O speed.c aa.c -o speed
$ time speed
2.01user 0.00system 0:02.00elapsed 100%CPU
In any case, you can prove it for yourself via the "strip" command, which strips the symbols from an executable.
This space intentionally left without comment.
--