OSS is not solely about cost. It's about freedoms. you may not want to develop OSS software yourself, but I'm sure you benefit from the interoperabilities and forks of OSS projects.
You choose free software because it gives you the freedom to use your computer the way you want.
It's very superficial to claim that the money savings is the only reason to switch to OSS from proprietary software.
I'm just expressing the fact that I see this as a win win in either case [Google wins and viacom dies] or [google loses and youtube dies].
I'm expressing that I have more important concerns in my life, and that I think people should demand a bit more from their media. Like news that is actually more objective than sensationalist. Shows that are [a combo of] silly, humourous, dramatic, informative, but not insulting the intelligence of the audience, etc.
Like for example, Law and Order. It has it's moments when they argue an interesting point of law, or chase the villain. But most of the episodes are tiresomely cliche and play to the more dominant emotions of the crowd (e.g. terrorism, rape, etc). Most sitcoms are based on the concept of sexual powers, as in the female almost always holds them all and the male goofs around in a vain attempt to get laid. There are only so many decades of "will he 'get some' or not?" sitcoms you can watch [or at least indirectly know about] before you just don't care anymore.
I'm not writing off all of pop culture. I'm just saying for the most part they're not even trying anymore. Survivor season 13? American Idol season 6? etc... Hint: most of the american idols this year are fantastically awful (no depth, no range). I'm surprised Simon hasn't jumped onto the stage and murdered one of them yet.
I just replaced Windows with Gentoo Linux. Worked better.:-)
Frankly I blame Intel for not having standard drivers on their f'ing website anywhere. The standard HDA driver works fine in Linux, why can't their be a standard HDA driver for WinXP?
Personally I'm really tired of Youtube. It's all that is wrong with insta-fame types. Though seeing people hurt themselves just to make it big on the intertubes is amusing...
And well, programming on TV hasn't really enthralled me since, well ever. The tired cliche shows may amuse the masses, good for them, but not anyone capable of doing a little thinking on their own.
Mine has been running fine [in Linux] for a year now. It seems like a solid design. Can't really complain about it.
In fact it works better in Linux than Windows as I can't find the damn Intel HDA drivers for this thing. Sound and wifi work out of the box with the Linux kernel.
My first laptop was a Compaq [forget the model] Athlon 1.8Ghz. It too was solid. I dropped it [in a metal case] twice, left it in the cold [in transit] and even poured about 100mL of water on it [by accident] and it kept going [after drying out].
In my experience Sony and Toshiba laptops are the flimsy'est even if the specs are otherwise nice. IBM laptops [lenovo now I guess] are also pretty tough though the ones I saw also weighed like 12lbs.
I'm not saying you can't do it (hint: I did it at one point).
I'm saying it's not fun for anyone else around. And if you plan on having friends or a SO in your life at anypoint, count this idea out as a long term plan.
Hmm top secret... could it be because everyone uses Office and it's proprietary formats? That's why.
No big conspiracy. People are just afraid of change and lazy. Same old same old. What I don't get is why people think you have to run the preloaded crap anyways. The first thing I did when I got my dell laptop was flash the HD and reinstall an OEM copy of windows. (well in the 2nd partition, Linux came first).
I suppose, but as others pointed out, this card isn't passively cooled. That seems like an oversight.
Anyways, I'm not entirely sold on the need to jump on the HD/BD bandwagon. They're still pressing DVDs. I'll wait till the last possible minute to upgrade [if I do at all]. Especially while they're still fighting over what resolutions to support... < 1080i == teh lame.
If my monitor can do 1280x1024 just fine, why can't a more expensive TV do 1080i or 1080p?
If you want to play video games on your TV, get a console. I can't imagine using a remote mouse/keyboard to play an FPS on a TV, especially if you have an SO or friends over. How boring.
IBM is full of bright people (and I don't mean that sarcastically). They are very smart. The problem is they hire a lot of fresh grads and give them write access to the main repository. Then they get them to work on some ancillary functionality before moving up. Unfortunately, for me, where my job was testing DB2 on different compilers, it's these ancillary functions that held me back.
That hash template for instance, took me about 2 months to fully work out why GCC didn't like it (it would fail at runtime), actually to this day I don't know exactly what was wrong, I just put a workaround in it's place. Worse yet, that hash template isn't even part of the majority of the runtime path, it's only called during startup/shutdown.
DB2 for instance, is the product of thousands of developers work. Most of whom are not even working on it anymore. So one guy may start a class [or method] and another add to it. But they might not take the design the same direction. So while the original developer may had one implementation in mind it ended up going another direction, the subsequent developers trying to wrestle what code there is to do what they think it should.
In the case of the hash template, I honestly think it was from a newb developer who didn't have the practical experience required to stop and think "there has to be a simpler solution." I remember when I was a newb developer (12-18 years old) I would write extremely long and complicated code for things that I could now accomplish with much more elegant code [and style]. These grads obviously didn't do much development during uni and the result is they honestly don't know how to express ideas concisely yet.
It's like if you're learning French [or any other second language]. You might use more verbose, or awkward language to express an idea the natives have simpler language for. For example, a newb may say "Nous parlons comme tout les autres Francais," whereas a French dude may say "Nous parlons comme du monde." The latter being more accepted and easily understood.
As for why management didn't catch it. Well several things. First, they're super busy at IBM. When I was there they were always running off to this meeting or that meeting. Second, it's not their job to sit and inspect the entire codebase (DB2 is also very large...). Third, even if they did see shitastic code, HR wouldn't let them fire/reassign them so easily.
They do have standards and testing suites. The problem is they're not comprehensive. At the time I started there, DB2 was only tested with ICC on x86 platforms. Even though they did support GCC (and used it on non-x86 platforms). Had they tested with GCC too, the hash template code would likely not have been accepted.
How do you plan to test those new comp.sci theories unless you can implement them?
How do you plan to maintain those implementations unless you have practical experience?
You can both develop software during your CS degree AND not consider it a waste of time. I'm sorry, but just because you know 30 different sorting algorithms, can multiply sparse matrices quickly, and design a stable neural network to play checkers, doesn't mean you can develop worth a damn. And very few places are interested in people who can't develop. In the practical world, comp.sci helps you become a better developer, in much the same an English course helps you communicate gooder.
I agree that many CS degrees are sidetracked by language issues. Java/C++/etc should be a semester long course. After that it should be expected that students will test their comp.sci knowledge in whatever language is appropriate, learning the syntax/etc as they go. Unfortunately, expecting the students to be responsible to research on their own is almost a crime nowadays...
There has to be some form of balance between the two. You can't do comp.sci with no programming (unless you don't plan to work for a living), and you can't do comp.sci by only focusing on the minute syntax details of a dozen languages.
Usually if you can cram functional code into tight spaces it's because you understand the algorithms that drive your program. That's why I made that comment. Someone who has no idea how sorting happens, relies on something like qsort, or whatever VB provides. Then when faced with that 16MHz ARM and 32KB of ram, they don't know what to do (other than copy/paste horrible code from the net).
And a computer scientist who can't develop software is about as useful as a rocket scientist who can't use a calculator [slide ruler, whatever]. Sometimes you have to get your hands dirty, especially since as far as I know there are no comp.sci farms where people just sit and dream up fancy math/algorithms (outside of uni that is).
10 Dim I as Integer 20 Put I into %eax why don't you? 30 Shift %eax by 4 bits to the left, please. 40 Store %eax back into I, hon.:-)
As for Java... I wonder why they all jump on that bandwagon. I have worked in a few shops [including AMD/IBM] and I have yet to actually be required to program something in Java.
I think we're making the same point more or less. To me, students should learn ASM and C first (bonus points for doing ASM on non-x86). But having a Java class in a comp.sci degree isn't a waste. Hell, most American unis put "Japanese History" as an undergrad requisit. It's all about balance.
You want to leave uni with a degree that says you know algorithm-fu and math-fu. But if you can't develop or code you're practically useless. Nobody hires "thinkers" to just dream up products. You need to actually code it. And a one semester class that introduces you to Java, Perl, Python, etc is not a bad idea.
When I was at IBM, I saw the product of "non-coder" developement. The result was "textbook" perfect code, that was totally inefficient and unmaintable. In one instance, I saw someone instantiate an entire template hashing class, to hash strings and sort them. He had about 3000 lines of code, full with nice non-descriptive block headers, for something that could have been written in 20 lines using qsort(). The code made used of all sorts of other things like a non-functional Montgomery reduction (to simulate the % operator for performance), inlined assembler, etc, etc, etc.
I should point out this hash class is called only during startup/shutdown and it's used to store internal PIDs to avoid clashes of processes. So the "performance need" is nil. Worst of all, it didn't build with GCC at all. Clearly the product of an overexcited developer who cracked open their Rivest Algorithms text and spewed out every idea they could. Clearly not the product of someone who has experience developing in C.
Do you know how much it costs IBM to maintain DB2 (the product I was assigned to)? A lot. Do you know how much it *should* cost them if it was coded properly? A lot less. IBM isn't special in this regard, I imagine the same thing happens at Oracle, at MSFT, at... I'm just picking on them since I have first hand experience.
So while I agree that a comp.sci degree should reflect a certain level of mathematics, algorithms, and the like, it should also reflect programming knowledge since that's what they end up doing anyways.
Does it have pointers? Can you drop into assembler with the ease of GCC? etc... Does it optimize on the same level as GCC or Sun CC? Just because it's compiled doesn't make it efficient. Both the language and the compilers have to have the capability of being efficient.
But that being said, you can still practice comp.sci in BASIC, that is the development of algorithms, datasets, etc. My comment about BASIC was how it was supposed to obsolete comp.sci, that "anyone can program it." Which may be somewhat true, doesn't mean they can produce what the world needs.
Think about it, if you have no idea about, say, numerical analysis, how could we ever hope to write something like MPEG? etc...
I think those are different directions though. At most a comp.sci degree should have a course on CVS/GIT, documentation, etc nothing more.
Should also point out that many different companies have different "standards" as far as development goes. So it's more important to understand the concepts than the actual fine point.
And frankly, nothing substitutes for experience. Comp.sci students should be encouraged to start OSS projects, maintain, and document them. If they started something like that in their 1st or 2nd year, by time they earn their masters they should already have a decent pile of experience from the real world.
I agree that comp.sci isn't strictly about programming, but I think it's a good idea to have *a* language in the courses. The problem otherwise is you have these "grads" who honestly don't know shit all about actually using their comp.sci knowledge. They end up having horribly unmaintainable coding practices that people like me usually have to come around and clean up.
Things like Java, C++, Perl, etc should be single semester long courses and not the focus of the degree. Because frankly, once you get the idea of programming languages, learning Java over C++ [or whatever] is a matter of the grammar really (especially since they're all fairly similar).
This has been asked repeatedly ever since I was a wee lad [20 years ago]. The idea then was BASIC would replace comp.sci because it was so simple to program. Of course, it overlooked the fact that BASIC is wickedly inefficient. No, the answer is no. No. No. No. Why? Someone's gotta maintain the scene.
For starters, the more automated tools are not efficient enough for most computing platforms (hint: think running that nice VB.NET application in 32KB of ram). Then combine that with the need for algorithms (re: 16MHz processors) and you can see that RAD tools don't apply.
If the memory is SRAM [or registers...] and hidden inside the IC, taking the chip apart to see where the memory is won't really help.
The idea is temporal security. In that, at some point the key goes over the bus [protected or otherwise] and cannot later be read back, that is, externally. Of course, inside the IC the memory is readable, how else would it use the key? But that's inside the IC with DPA/SPA resistance and the like...
There is a whole build up for "keywrap" standards which address this very problem. E.g. RSA or ECC encrypt the key, fire it over the bus, the IC decrypts it [with a private key stored internally]. The key could be encrypted on the host processor or even externally (e.g. to authenticate something).
How is that a troll? i work for a hardware firm. We do this on a regular basis. People want to be able to feed a key to something and not have bus snoop read it later.
I only called the OP a newb because he/she/it was being all sarcastic about something we do on a regular basis.
Slashdot Mirror
OSS is not solely about cost. It's about freedoms. you may not want to develop OSS software yourself, but I'm sure you benefit from the interoperabilities and forks of OSS projects.
You choose free software because it gives you the freedom to use your computer the way you want.
It's very superficial to claim that the money savings is the only reason to switch to OSS from proprietary software.
Tom
I'm just expressing the fact that I see this as a win win in either case [Google wins and viacom dies] or [google loses and youtube dies].
I'm expressing that I have more important concerns in my life, and that I think people should demand a bit more from their media. Like news that is actually more objective than sensationalist. Shows that are [a combo of] silly, humourous, dramatic, informative, but not insulting the intelligence of the audience, etc.
Like for example, Law and Order. It has it's moments when they argue an interesting point of law, or chase the villain. But most of the episodes are tiresomely cliche and play to the more dominant emotions of the crowd (e.g. terrorism, rape, etc). Most sitcoms are based on the concept of sexual powers, as in the female almost always holds them all and the male goofs around in a vain attempt to get laid. There are only so many decades of "will he 'get some' or not?" sitcoms you can watch [or at least indirectly know about] before you just don't care anymore.
I'm not writing off all of pop culture. I'm just saying for the most part they're not even trying anymore. Survivor season 13? American Idol season 6? etc... Hint: most of the american idols this year are fantastically awful (no depth, no range). I'm surprised Simon hasn't jumped onto the stage and murdered one of them yet.
Tom
I just replaced Windows with Gentoo Linux. Worked better. :-)
Frankly I blame Intel for not having standard drivers on their f'ing website anywhere. The standard HDA driver works fine in Linux, why can't their be a standard HDA driver for WinXP?
Tom
A win win.
Personally I'm really tired of Youtube. It's all that is wrong with insta-fame types. Though seeing people hurt themselves just to make it big on the intertubes is amusing...
And well, programming on TV hasn't really enthralled me since, well ever. The tired cliche shows may amuse the masses, good for them, but not anyone capable of doing a little thinking on their own.
Tom
Mine has been running fine [in Linux] for a year now. It seems like a solid design. Can't really complain about it.
In fact it works better in Linux than Windows as I can't find the damn Intel HDA drivers for this thing. Sound and wifi work out of the box with the Linux kernel.
My first laptop was a Compaq [forget the model] Athlon 1.8Ghz. It too was solid. I dropped it [in a metal case] twice, left it in the cold [in transit] and even poured about 100mL of water on it [by accident] and it kept going [after drying out].
In my experience Sony and Toshiba laptops are the flimsy'est even if the specs are otherwise nice. IBM laptops [lenovo now I guess] are also pretty tough though the ones I saw also weighed like 12lbs.
Tom
I'm not saying you can't do it (hint: I did it at one point).
I'm saying it's not fun for anyone else around. And if you plan on having friends or a SO in your life at anypoint, count this idea out as a long term plan.
Tom
Hmm top secret ... could it be because everyone uses Office and it's proprietary formats? That's why.
No big conspiracy. People are just afraid of change and lazy. Same old same old. What I don't get is why people think you have to run the preloaded crap anyways. The first thing I did when I got my dell laptop was flash the HD and reinstall an OEM copy of windows. (well in the 2nd partition, Linux came first).
tom
Darn you!!! hehehehe I came here to say the same thing (right down to the batman paraphrase).
I suppose, but as others pointed out, this card isn't passively cooled. That seems like an oversight.
Anyways, I'm not entirely sold on the need to jump on the HD/BD bandwagon. They're still pressing DVDs. I'll wait till the last possible minute to upgrade [if I do at all]. Especially while they're still fighting over what resolutions to support... < 1080i == teh lame.
If my monitor can do 1280x1024 just fine, why can't a more expensive TV do 1080i or 1080p?
Tom
If you want to play video games on your TV, get a console. I can't imagine using a remote mouse/keyboard to play an FPS on a TV, especially if you have an SO or friends over. How boring.
:-)
Get a Wii
Tom
Wait, I still use a VGA monitor, with a higher dot pitch than most any HD TV ...
I guess this is good for folks who build home theatres out of their computers, but then why do they need a 3D accelerator to show TV or videos?
Tom
IBM is full of bright people (and I don't mean that sarcastically). They are very smart. The problem is they hire a lot of fresh grads and give them write access to the main repository. Then they get them to work on some ancillary functionality before moving up. Unfortunately, for me, where my job was testing DB2 on different compilers, it's these ancillary functions that held me back.
...). Third, even if they did see shitastic code, HR wouldn't let them fire/reassign them so easily.
That hash template for instance, took me about 2 months to fully work out why GCC didn't like it (it would fail at runtime), actually to this day I don't know exactly what was wrong, I just put a workaround in it's place. Worse yet, that hash template isn't even part of the majority of the runtime path, it's only called during startup/shutdown.
DB2 for instance, is the product of thousands of developers work. Most of whom are not even working on it anymore. So one guy may start a class [or method] and another add to it. But they might not take the design the same direction. So while the original developer may had one implementation in mind it ended up going another direction, the subsequent developers trying to wrestle what code there is to do what they think it should.
In the case of the hash template, I honestly think it was from a newb developer who didn't have the practical experience required to stop and think "there has to be a simpler solution." I remember when I was a newb developer (12-18 years old) I would write extremely long and complicated code for things that I could now accomplish with much more elegant code [and style]. These grads obviously didn't do much development during uni and the result is they honestly don't know how to express ideas concisely yet.
It's like if you're learning French [or any other second language]. You might use more verbose, or awkward language to express an idea the natives have simpler language for. For example, a newb may say "Nous parlons comme tout les autres Francais," whereas a French dude may say "Nous parlons comme du monde." The latter being more accepted and easily understood.
As for why management didn't catch it. Well several things. First, they're super busy at IBM. When I was there they were always running off to this meeting or that meeting. Second, it's not their job to sit and inspect the entire codebase (DB2 is also very large
They do have standards and testing suites. The problem is they're not comprehensive. At the time I started there, DB2 was only tested with ICC on x86 platforms. Even though they did support GCC (and used it on non-x86 platforms). Had they tested with GCC too, the hash template code would likely not have been accepted.
Anyways nuff ranting...
How do you plan to test those new comp.sci theories unless you can implement them?
How do you plan to maintain those implementations unless you have practical experience?
You can both develop software during your CS degree AND not consider it a waste of time. I'm sorry, but just because you know 30 different sorting algorithms, can multiply sparse matrices quickly, and design a stable neural network to play checkers, doesn't mean you can develop worth a damn. And very few places are interested in people who can't develop. In the practical world, comp.sci helps you become a better developer, in much the same an English course helps you communicate gooder.
I agree that many CS degrees are sidetracked by language issues. Java/C++/etc should be a semester long course. After that it should be expected that students will test their comp.sci knowledge in whatever language is appropriate, learning the syntax/etc as they go. Unfortunately, expecting the students to be responsible to research on their own is almost a crime nowadays...
There has to be some form of balance between the two. You can't do comp.sci with no programming (unless you don't plan to work for a living), and you can't do comp.sci by only focusing on the minute syntax details of a dozen languages.
Tom
Usually if you can cram functional code into tight spaces it's because you understand the algorithms that drive your program. That's why I made that comment. Someone who has no idea how sorting happens, relies on something like qsort, or whatever VB provides. Then when faced with that 16MHz ARM and 32KB of ram, they don't know what to do (other than copy/paste horrible code from the net).
And a computer scientist who can't develop software is about as useful as a rocket scientist who can't use a calculator [slide ruler, whatever]. Sometimes you have to get your hands dirty, especially since as far as I know there are no comp.sci farms where people just sit and dream up fancy math/algorithms (outside of uni that is).
Assembler in BASIC? That's scary.
:-)
... I wonder why they all jump on that bandwagon. I have worked in a few shops [including AMD/IBM] and I have yet to actually be required to program something in Java.
10 Dim I as Integer
20 Put I into %eax why don't you?
30 Shift %eax by 4 bits to the left, please.
40 Store %eax back into I, hon.
As for Java
Tom
I think we're making the same point more or less. To me, students should learn ASM and C first (bonus points for doing ASM on non-x86). But having a Java class in a comp.sci degree isn't a waste. Hell, most American unis put "Japanese History" as an undergrad requisit. It's all about balance.
... I'm just picking on them since I have first hand experience.
You want to leave uni with a degree that says you know algorithm-fu and math-fu. But if you can't develop or code you're practically useless. Nobody hires "thinkers" to just dream up products. You need to actually code it. And a one semester class that introduces you to Java, Perl, Python, etc is not a bad idea.
When I was at IBM, I saw the product of "non-coder" developement. The result was "textbook" perfect code, that was totally inefficient and unmaintable. In one instance, I saw someone instantiate an entire template hashing class, to hash strings and sort them. He had about 3000 lines of code, full with nice non-descriptive block headers, for something that could have been written in 20 lines using qsort(). The code made used of all sorts of other things like a non-functional Montgomery reduction (to simulate the % operator for performance), inlined assembler, etc, etc, etc.
I should point out this hash class is called only during startup/shutdown and it's used to store internal PIDs to avoid clashes of processes. So the "performance need" is nil. Worst of all, it didn't build with GCC at all. Clearly the product of an overexcited developer who cracked open their Rivest Algorithms text and spewed out every idea they could. Clearly not the product of someone who has experience developing in C.
Do you know how much it costs IBM to maintain DB2 (the product I was assigned to)? A lot. Do you know how much it *should* cost them if it was coded properly? A lot less. IBM isn't special in this regard, I imagine the same thing happens at Oracle, at MSFT, at
So while I agree that a comp.sci degree should reflect a certain level of mathematics, algorithms, and the like, it should also reflect programming knowledge since that's what they end up doing anyways.
Tom
Does it have pointers? Can you drop into assembler with the ease of GCC? etc... Does it optimize on the same level as GCC or Sun CC? Just because it's compiled doesn't make it efficient. Both the language and the compilers have to have the capability of being efficient.
But that being said, you can still practice comp.sci in BASIC, that is the development of algorithms, datasets, etc. My comment about BASIC was how it was supposed to obsolete comp.sci, that "anyone can program it." Which may be somewhat true, doesn't mean they can produce what the world needs.
Think about it, if you have no idea about, say, numerical analysis, how could we ever hope to write something like MPEG? etc...
Tom
I think those are different directions though. At most a comp.sci degree should have a course on CVS/GIT, documentation, etc nothing more.
Should also point out that many different companies have different "standards" as far as development goes. So it's more important to understand the concepts than the actual fine point.
And frankly, nothing substitutes for experience. Comp.sci students should be encouraged to start OSS projects, maintain, and document them. If they started something like that in their 1st or 2nd year, by time they earn their masters they should already have a decent pile of experience from the real world.
I agree that comp.sci isn't strictly about programming, but I think it's a good idea to have *a* language in the courses. The problem otherwise is you have these "grads" who honestly don't know shit all about actually using their comp.sci knowledge. They end up having horribly unmaintainable coding practices that people like me usually have to come around and clean up.
Things like Java, C++, Perl, etc should be single semester long courses and not the focus of the degree. Because frankly, once you get the idea of programming languages, learning Java over C++ [or whatever] is a matter of the grammar really (especially since they're all fairly similar).
Tom
This has been asked repeatedly ever since I was a wee lad [20 years ago]. The idea then was BASIC would replace comp.sci because it was so simple to program. Of course, it overlooked the fact that BASIC is wickedly inefficient. No, the answer is no. No. No. No. Why? Someone's gotta maintain the scene.
For starters, the more automated tools are not efficient enough for most computing platforms (hint: think running that nice VB.NET application in 32KB of ram). Then combine that with the need for algorithms (re: 16MHz processors) and you can see that RAD tools don't apply.
Tom
If the memory is SRAM [or registers...] and hidden inside the IC, taking the chip apart to see where the memory is won't really help.
The idea is temporal security. In that, at some point the key goes over the bus [protected or otherwise] and cannot later be read back, that is, externally. Of course, inside the IC the memory is readable, how else would it use the key? But that's inside the IC with DPA/SPA resistance and the like...
There is a whole build up for "keywrap" standards which address this very problem. E.g. RSA or ECC encrypt the key, fire it over the bus, the IC decrypts it [with a private key stored internally]. The key could be encrypted on the host processor or even externally (e.g. to authenticate something).
Tom
I wouldn't want users installing applications anywhere but in their own directories. That's what makes Windows so damn insecure in the first place.
There already quite a few random GUI installers for Linux though. UT2K4 [the game] uses one. So do Samsung laser printer drivers.
Tom
How is that a troll? i work for a hardware firm. We do this on a regular basis. People want to be able to feed a key to something and not have bus snoop read it later.
I only called the OP a newb because he/she/it was being all sarcastic about something we do on a regular basis.
Um, it exists. Basically you put memory behind a controller which does not allow reads from a given bus. Hence, write only.
NEWBIE!
But CBC requires IVs. Are they using up sectors to store them?
The whole idea of XTS is that you can get privacy without extra storage.
Tom