Was this a one time thing, or could you do this repeatedly? If you made one performance tweak and accurately saw improvement, that doesn't say a whole lot. After all, you were expecting some improvement to begin with. What would be more conclusive is if you never perceived any improvement in times where you made an adjustment and there was no improvement.
I'm not a psychologist, but I'm pretty sure that that 5%-10% changes you mentioned are pretty close to the threshold of detection for most people. So most people might perceive a lot of changes correctly, but might also make errors if forced to determine whether a change has really occurred or not. I'm think that quite a few studies have been done to determine various human sensitivity thresholds. YMMV, but its tough to say for sure without a proper experiment to test it.
Hmmm, I've had my own problems with trying to change application defaults. For me the main thing has been trying to get my install on Debian which originally came from Knoppix to use Letter size paper instead of A4. I have to reset this for every document (if anyone knows how to change this please reply).
As far as style defaults, I haven't tried it but you might want to look into defining a template from an existing document with the appropriately modified styles. I think the idea is that people will tend to want different styles for different types of documents, so there isn't much sense in allowing modifications to the defaults styles for everything. A template for each type of document you want to write makes more sense, and shouldn't be much work. If nothing else, just take an existing document with updated styles, delete all the content, and there you have a base to start new documents. I think that this is essentially what an MS Word template is anyways, though don't quote me on that.
The look and feel of the program is a bit too rough. For example, they inexplicably have a huge "Styles" pane on by default that covers 1/4 of the document.
This is something I really like, because it helps show users the correct way to format their documents.
I haven't used MS Word for years, and I never really have done serious word processing with it, but I never knew about styles, or stylesheets or whatever they're called in Word. I always formatted each paragraph that needed to conform to some style on it's own by setting paragraph settings, etc.
With OpenOffice it is much clearer how to do proper formatting. Choose the appropriate type for each part of a document (header, title, main body), and just modify the style for that type once through the stylist dialog. I'm guessing that this is basically how it should be done in Word as well, but I never delved deeply into it enough to figure it out. With OpenOffice they point you in the right direction. Learning to use the stylist is a great time and effort saver in the medium to long term, so I think it's great that it's made obvious to new users.
I could notice speed differences between 5%-10%. (Like RAM got a 7% speed increase because of the latest bios tweak - I could "feel" it even before win(2k) was booted. It's weird, but true)
I don't buy it. Your brain wants to believe that every little tweak you performed was beneficial and actually worth the time you spent putting it into place. Even if the tweak actually slowed down your system by as much as 5% you'd probably not notice it, and might even subconsciously believe it was faster to validate your efforts.
I'm sure that many tweaks did actually work, but if you actually conducted a double blind study to truly test whether someone could tell if their computer is running 3% slower (reasonable amount for an efficient virus) I think you'd see that nearly as many people incorrectly think the computer was running faster as correctly perceived it to run slower. People just aren't that sensitive to such small deviations, especially if they aren't looking for the in the first place, which most people aren't.
Will people stop modding these idiots up? Having a low Slashdot ID means nothing. Look at their names, none of these guys regularly contribute much of anything anymore, if they ever did. All they do is lurk and post whenever someone makes a crack about having a low ID number.
I mean, it was funny the first couple times it happened, but this type of thread comes up on a near weekly basis now. These guys are nothing but elitist karma whores. I say treat them as such.
It amounts to rendering images of your problem, then doing feature extraction on the image.
This may be one approach, but it is certainly not the only one. The Brook software (linked a few replies above) takes advantage of modern, programmable GPUs to write programs that have little to do with graphics rendering. I saw a talk by Pat Hanrahan, the guy leading the project, and for some mathematical algorithms they achieved something like 10x speedup over highly optimized, best of breed implementations compiled with Intel's C compiler.
There are issues, possibly the biggest being that GPU programs are limited in size and the GPU bandwidth on the AGP bus is assymetric.
Uh, maybe you missed the "heh" at the beginning of the last part, but it was a tounge-in-cheek statement. I know that abolishing copyright is not what (almost) any interested party really wants, but ironically it would remove the ability for people to use the GPL as the "restraint of trade" the bozos mentioned are complaining about. People would simply be able to do the same for any work.
The bozos behing the actions in the article would very much like to see the GPL stripped of the powers you mention, which is why I posted abolishing copyright as a "solution" that no one really wants. It would fulfill the desire to be able to do anything a person wanted with GPL code, but at the price of a level playing field for all code.
the GPL is nothing more than a price fixing scheme
Actually, it's not the GPL which is a price fixing scheme, but copyright. As in copyright allows the author to set any price they want to license their work. Somehow I don't think Wallace, O'Gara, and countless others quite understand this simple fact (or are not willing to admit it, as this pseudo-monopolistic characteristic of copyright seems to fly in the face of other free market ideals).
The GPL actually removes this restriction by allowing a copyrighted work to be licensed for any amount of money desired by any party with a copy of the work. Zero just happens to be the most typical number, for the practical reason that it is difficult to get customers to pay exorbitant sums for what is usually available elsewhere for free.
Heh, maybe we should just abolish copyright and remove this restraint of trade in all cases. This way anyone could license any work created by anyone else for however much they thought they could get, though this probably isn't the type of "solution" the software industry has in mind.
I realize that's the point, but has it actually produced any results? Any theorems, validated user tests? I don't know many details about it other than that Jef Raskin supposedly worked on this thing or ideas like it for something like the past 20 years, but seems to have very little to actually show for it. When someone rails against the establishment for that long, they better actually produce something.
As far as Raskin being a "god", I haven't seen it. He had more than enought time to prove the viability of his ideas. Why is it that we hardly see anything that actually works until after he dies? My guess is that any prototypes he's made were flawed in some way, and he's never been able to correct them such that they could pass muster in the public. Something was likely missing in the transition from good sounding idea to practical application.
In any case, if they ever make a version of Archy that runs on Linux I'll be glad to test it. It should be trivial to make it cross platform since it doesn't sound like ti would rely on any standard widgets for drawing, although I suppose the ridiculous keymapping and USB keyboard extension requirements would obviously make it more difficult.
Care to prove this? I'm serious. If you're really advocating dismissing the majority of existing interfaces for something that has never moved beyond the laboratory/prototype phase you'd better have some way of backing these statements up.
Are there theorems of human-computer interaction efficiency that prove that Archy is more efficient than WIMP interfaces? Have extensive user studies been conducted with test subjects having wide ranges of computer experience?
I'm not trying to rain on your parade here, but lots of people make grandiose claims about revolutionizing just about any field you can think of. Very rarely do these claims amount to much, in most cases because no matter how good an idea seems in theory there usually exists one or more flaws that bring it down in practice. It's up to Archy's developers and supporters to demonstrate otherwise with hard evidence, not just compelling ideas.
Above AC should not be marked troll. The statement is precisely true. The code is not the most important thing. Academic publications are about ideas and methods, and specifically is CS, algorithms.
I'm a CS masters student. I'm taking a class in Computational Geometry. How much code has been presented to me in the class? None. Does this mean that I'm not getting a quality education? Hardly. The combination of pseudocode, description, theorems and proofs are far more useful for understanding a problem in whole than some single implementation would be. Implementations usually include details that are of little interest and would clutter the issue, obscuring the important bits.
Often, code created for research is not really production quality anyways, it is written specifically for a single problem. When code is high quality and sufficiently general purpose that it would actually be useful to others it often is released. In my field of computer graphics, code like this has been released which was used in writing several papers I have read and used.
While there is certainly some politics in CS publication, I doubt there is significantly more than any other academic field, and likely a good bit less than many "softer" fields.
You are correct that code is rarely published, that code quality may be poor, and that said code is probably not reusable. I say you're missing the point, the goal of CS is to create or discover new algorithms and methods, not write production quality code. If that's what you want I suggest you study software engineering (a branch of CS by most groupings, but with significantly different goals than most other areas). In any case, creating finished products is NOT the goal of academic research, and CS should not be an exception to this rule.
As far as not being verifiable, I say BS. The methods described in most CS articles easily give enough information for a person skilled and knowledgable in the field to duplicate the work. I've personally implemented a few fairly complex algorithms based only on what is described in a research paper. Additionally, duplicating a CS method takes considerably less effort than duplicating most experiments in the physical sciences. Everyone has access to computers, not many people have access to things like giant telescopes or particle accelerators, and the only resources consumed in algorithm implementation are time and electricity.
In conclusion, I'm not terribly worried about CS research. Lots of great projects are being done that fill both theoretical and practical voids. These ideas will filter through to practical applications and/or inspire more new ideas, and progress will march on.
I can eat fast food for $20/week. No matter how I run the numbers, I can't eat truely healthy for less than $100/week.
Your kidding right? My food budget is about $50-$60 per week, and it could probably be less if I cooked more from scratch (I do cook most of my own food, but there are usually a few shortcuts compared to completely from scratch). I eat a lot of salads, which are healthy and not expensive. I eat more chicken than beef, which is both a bit healthier and fairly inexpensive.
Maybe that's not that much less than $100, but there's also no way you get by on $20. 2 Big Macs per day? That's what, $7 for the value meals and at least $4 just for the burgers? Already at about $30 per wekk or more.
In any case, I think that your health (not to mention the pleasure of eating real food instead of crap) is worh an extra few bucks a day. If not, then your budget likely needs some prioritizing, as you probably waste more money elsewhere (not because you specifically can't budget, but because almost everyone has some waste).
You bring up an interesting point, but there are a few other factors to consider. You see that 40-60 hours as time down the tubes, but was it really? A few thoughts:
He's new, so much of that week was probably also spent getting acclimated to the company, it's systems, and its people, which would have taken some time in any case.
Maybe he's more productive in Linux. If that's what he uses best, then even if Linux only saved him one hour per week over OS X the Linux machine would pay back its setup costs in one year.
Even if he would have eventually become more productive with a Mac, if he's not used to it there would be an initial learning time cost, even for a computer expert.
Assuming an employee costs a company $100K/year (salary, benefits, utilities and other resources), one week would have covered about one powerbook.
I don't know what the best decision really would have been, but just like TCO analysis you can't get the big picture by only looking at one factor.
Interesting. So you complain about how CC signatures are basically useless (no argument there), and then go on and bitch because some stores ask for a photo ID?
You do realize that a photo ID is much more effective in stopping CC abuse than a signature, don't you? As in most potential CC thieves don't look much like you, even if they could duplicate your signature (or get away with any old signature)?
Marketing means finding out what people want, finding out what people need and then making it and getting it sold.
Actually, what you're describing here is market research. Marketing is find ways to get people to buy a product that has already been created. Market research tends to benefit both producer and consumer by allowing creation of better products. Marketing has more one-sided benefits in favor of producers, as good marketing might be getting lots of people to buy an inferior or overpriced product.
I've used Mozilla for mail for the past several years, and I've had no problem trasnferring my POP folders to a new machine. I don't get all that much mail either, so the storage space wouldn't even be an issue, except...
Attachments that I no longer care about take up I'm sure the majority of the storage. I'd like to be able to delete the attachments while keeping the messages. Does anyone know of any tools that can go through mail in POP3 format and allow selective deleting of attachments. Ideally I'd like it to be a separate tool from an actual email program, but if there's an email program that can do this (and runs on Linux) I might consider setting it up on my existing folders just to allow deleting of attachments on occasion, while uisng Mozilla to actually download, send and read email.
Re:Is it any wonder I didn't get laid in Universit
on
Women Leaving I.T.
·
· Score: 1
I wouldn't recommend computer science to my worst enemy.
Not only do you not meet any girls in any of your classes you carry around the stigma of comp sci.
Maybe your problem is that you only spent time with CS people.
I had a great time as an undergrad in CS. Did lots of partying (esp. living in the dorms freshman and sophomore years), played intramural flag football, and did well academically to boot. None of my friends were in CS though, mostly I met them through living in the dorms or something like that. This might be more difficult if you go to a completely tech oriented school, but at a more balanced school with large departments in the humanities and such it shouldn't be a big deal.
As long as your fairly self sufficient academically there's no real reason to heng out with people in your major any more than you want. Building contacts for future jobs maybe (although apparently even this didn't work out to well for you!), but that's about it.
I originally had you pegged for a troll from your initial snide remark, maybe I was wrong, we may actually see more eye to eye than I would have thought on this one. Still, I don't think it is a question of having it both ways. There are plently of people who feel corps have far to much power to control copyright who don't necessarily want to make unauthorized copies of stuff purely because they're too cheap to pay for it. If anything the corps are the ones who want to have it both ways, they want the benefits of government granted copyright protection without the trade off of eventual release into public domain, or enabling any semblance of fair use.
Identity vandalism is a somewhat better term than identity theft, but doesn't really convey the seriousness of the crime, and really there are elements of theft, vandalism, and outright fraud. It's like theft because their is only one identity, and it can be used by others for unauthorized purposes which benefit the identity thief (unlike pure vandalism). It is also like vandalism because real damage is caused by the time you get control of your identity back. It bears very little resemblance to copyright infringment (or piracy, or whatever term you want to use), though, which is what prompted my initial remark in the first place.
Nice try but no, jackass. Copyright infringement is not theft because me making a copy of some data does not alter the copy that you or anyone else has in their possession.
Identity theft is much closer to physical theft because their is only one record of a person's identity in any of these databases. When someone steals your identity it does not create a new copy in the database, it fucks up your one and only copy. Just like physical theft and abuse of an item fucks up the one one and only instance of the particular item that was stolen.
Maybe theft isn't the perfect term to describe it, but identity theft bears a lot more resemblance to physical property theft than unauthorized data copying does.
Yes, but the parent was stating a difference between apps which require 3D capabilities themselves vs. those that only require 3D like capabilities from the window manager.
2D apps treated like textured polys do not need their own depth buffer, frame buffer, etc. They just need a texture buffer, and the window manager treats them like texture polys in a single, comprehensive 3D app. Conventional 3D apps require their own depth buffer, frame buffer, textures, etc. in addition to that used by the window manager. I was pointing out that while the requirements for 30+ 2D apps are fairly large, they are considereably smaller than the same number of true 3D apps, because they have needs beyond the WM overhead.
All in all this is a pretty fair assessment, but I think it leans a bit to the pessemistic.
First, I doubt a window manager would actually use a 32 bit z-buffer. 8 bit would be overkill here (enought to specify a unique depth for 256 windows). Even a 3D window manager would get by on 16 bit depth no problem, I believe it's the most commonly used depth for most true 3D apps now.
Also, I doubt that in many cases more than a small number of true 3D windows would be needed. Someone who is working with 30+ windows open most likely has mostly terminals, web pages and text editors open, with maybe a few 3D apps.
Sorry, while I agree with the general thrust of this your 10 line rule seems overly pedantic and ultimately counterproductive.
Often I have written largeish functions such that while they could be broken into smaller functions, at the time there would be no purpose in doing so except to enforce the "only short functions allowed" rule. In other words the small functions would only be called from a single point in the code. In situations like this recursively breaking a function into smaller functions adds function call overhead which impacts performance, but more importantly adds additional whitespace and boilerplate function definition code to the source that can detract from readability. This is especially true if a developer has to mentally step through every line of code in a function: if the function is monolithic the step through is basically linear, while if the function is heavily subdivided the developer must jump back and forth in the code to follow the chain of execution.
Yes, yes, ideally each function should be tested and verified individually, but if a function was custom coded for a specific purpose only to be used in one larger function then this is often difficult to do, due to lack of alternative test situations.
It is often the case that a chunk of code that is only useful in one location early on may be useful in multiple places later, but I believe it is best to wait until this time to refactor the original code into smaller, more modular functions.
The flat line they are talking about is likely a rolling average. With a rolling average when the results are trully random the display will be flat. However when the results break away from randomness the average will tend away from the midline.
This would be the easiest way to graph deviations from true randomness.
Re:Maybe you should read a book / the spec
on
The CSS Anthology
·
· Score: 1
The best reason is if you ever want to change the layout of your site in the future.
If done correctly and with foresight you can design complete a CSS layout which can be completely modified at a later date without a single change to your (X)HTML code. This isn't trivial, as it is hard to know for certain what changes you may want in the future.
The typicaly scheme which I use on my rather simply laid out website (monsterden.net) is basically to divide the code up into chunks using divs, and assign IDs or classes to divs based on content type. Doing this you will generally only run into problems if you decide at a later time that you want to rearrange blocks with finer granularity than you initially planned.
Another reason which does not apply in all cases is readability. If you look a the code on my site you'll see that it is extremely readable. This is important to me because I code my site completely by hand. CSS removes a lot of table clutter and makes nesting of code blocks less of an issue.
Finally you can do some rather nifty tricks with CSS that would otherwise require Javascript. On my background images pages I use a few of these to create some nice artistic effects, including the changing of the cneter image when a link is hovered over (NOTE: I'm not sure if the CSS on this particular part of my site works in MSIE, as I went for coolness over full compatability here. The rest of my site is fully compatible with all modern browsers AKAIK).
Slashdot Mirror
Was this a one time thing, or could you do this repeatedly? If you made one performance tweak and accurately saw improvement, that doesn't say a whole lot. After all, you were expecting some improvement to begin with. What would be more conclusive is if you never perceived any improvement in times where you made an adjustment and there was no improvement.
I'm not a psychologist, but I'm pretty sure that that 5%-10% changes you mentioned are pretty close to the threshold of detection for most people. So most people might perceive a lot of changes correctly, but might also make errors if forced to determine whether a change has really occurred or not. I'm think that quite a few studies have been done to determine various human sensitivity thresholds. YMMV, but its tough to say for sure without a proper experiment to test it.
Hmmm, I've had my own problems with trying to change application defaults. For me the main thing has been trying to get my install on Debian which originally came from Knoppix to use Letter size paper instead of A4. I have to reset this for every document (if anyone knows how to change this please reply).
As far as style defaults, I haven't tried it but you might want to look into defining a template from an existing document with the appropriately modified styles. I think the idea is that people will tend to want different styles for different types of documents, so there isn't much sense in allowing modifications to the defaults styles for everything. A template for each type of document you want to write makes more sense, and shouldn't be much work. If nothing else, just take an existing document with updated styles, delete all the content, and there you have a base to start new documents. I think that this is essentially what an MS Word template is anyways, though don't quote me on that.
The look and feel of the program is a bit too rough. For example, they inexplicably have a huge "Styles" pane on by default that covers 1/4 of the document.
This is something I really like, because it helps show users the correct way to format their documents.
I haven't used MS Word for years, and I never really have done serious word processing with it, but I never knew about styles, or stylesheets or whatever they're called in Word. I always formatted each paragraph that needed to conform to some style on it's own by setting paragraph settings, etc.
With OpenOffice it is much clearer how to do proper formatting. Choose the appropriate type for each part of a document (header, title, main body), and just modify the style for that type once through the stylist dialog. I'm guessing that this is basically how it should be done in Word as well, but I never delved deeply into it enough to figure it out. With OpenOffice they point you in the right direction. Learning to use the stylist is a great time and effort saver in the medium to long term, so I think it's great that it's made obvious to new users.
I could notice speed differences between 5%-10%. (Like RAM got a 7% speed increase because of the latest bios tweak - I could "feel" it even before win(2k) was booted. It's weird, but true)
I don't buy it. Your brain wants to believe that every little tweak you performed was beneficial and actually worth the time you spent putting it into place. Even if the tweak actually slowed down your system by as much as 5% you'd probably not notice it, and might even subconsciously believe it was faster to validate your efforts.
I'm sure that many tweaks did actually work, but if you actually conducted a double blind study to truly test whether someone could tell if their computer is running 3% slower (reasonable amount for an efficient virus) I think you'd see that nearly as many people incorrectly think the computer was running faster as correctly perceived it to run slower. People just aren't that sensitive to such small deviations, especially if they aren't looking for the in the first place, which most people aren't.
Will people stop modding these idiots up? Having a low Slashdot ID means nothing. Look at their names, none of these guys regularly contribute much of anything anymore, if they ever did. All they do is lurk and post whenever someone makes a crack about having a low ID number.
I mean, it was funny the first couple times it happened, but this type of thread comes up on a near weekly basis now. These guys are nothing but elitist karma whores. I say treat them as such.
It amounts to rendering images of your problem, then doing feature extraction on the image.
This may be one approach, but it is certainly not the only one. The Brook software (linked a few replies above) takes advantage of modern, programmable GPUs to write programs that have little to do with graphics rendering. I saw a talk by Pat Hanrahan, the guy leading the project, and for some mathematical algorithms they achieved something like 10x speedup over highly optimized, best of breed implementations compiled with Intel's C compiler.
There are issues, possibly the biggest being that GPU programs are limited in size and the GPU bandwidth on the AGP bus is assymetric.
Uh, maybe you missed the "heh" at the beginning of the last part, but it was a tounge-in-cheek statement. I know that abolishing copyright is not what (almost) any interested party really wants, but ironically it would remove the ability for people to use the GPL as the "restraint of trade" the bozos mentioned are complaining about. People would simply be able to do the same for any work.
The bozos behing the actions in the article would very much like to see the GPL stripped of the powers you mention, which is why I posted abolishing copyright as a "solution" that no one really wants. It would fulfill the desire to be able to do anything a person wanted with GPL code, but at the price of a level playing field for all code.
the GPL is nothing more than a price fixing scheme
Actually, it's not the GPL which is a price fixing scheme, but copyright. As in copyright allows the author to set any price they want to license their work. Somehow I don't think Wallace, O'Gara, and countless others quite understand this simple fact (or are not willing to admit it, as this pseudo-monopolistic characteristic of copyright seems to fly in the face of other free market ideals).
The GPL actually removes this restriction by allowing a copyrighted work to be licensed for any amount of money desired by any party with a copy of the work. Zero just happens to be the most typical number, for the practical reason that it is difficult to get customers to pay exorbitant sums for what is usually available elsewhere for free.
Heh, maybe we should just abolish copyright and remove this restraint of trade in all cases. This way anyone could license any work created by anyone else for however much they thought they could get, though this probably isn't the type of "solution" the software industry has in mind.
I realize that's the point, but has it actually produced any results? Any theorems, validated user tests? I don't know many details about it other than that Jef Raskin supposedly worked on this thing or ideas like it for something like the past 20 years, but seems to have very little to actually show for it. When someone rails against the establishment for that long, they better actually produce something.
As far as Raskin being a "god", I haven't seen it. He had more than enought time to prove the viability of his ideas. Why is it that we hardly see anything that actually works until after he dies? My guess is that any prototypes he's made were flawed in some way, and he's never been able to correct them such that they could pass muster in the public. Something was likely missing in the transition from good sounding idea to practical application.
In any case, if they ever make a version of Archy that runs on Linux I'll be glad to test it. It should be trivial to make it cross platform since it doesn't sound like ti would rely on any standard widgets for drawing, although I suppose the ridiculous keymapping and USB keyboard extension requirements would obviously make it more difficult.
One method must be better than another.
Care to prove this? I'm serious. If you're really advocating dismissing the majority of existing interfaces for something that has never moved beyond the laboratory/prototype phase you'd better have some way of backing these statements up.
Are there theorems of human-computer interaction efficiency that prove that Archy is more efficient than WIMP interfaces? Have extensive user studies been conducted with test subjects having wide ranges of computer experience?
I'm not trying to rain on your parade here, but lots of people make grandiose claims about revolutionizing just about any field you can think of. Very rarely do these claims amount to much, in most cases because no matter how good an idea seems in theory there usually exists one or more flaws that bring it down in practice. It's up to Archy's developers and supporters to demonstrate otherwise with hard evidence, not just compelling ideas.
Above AC should not be marked troll. The statement is precisely true. The code is not the most important thing. Academic publications are about ideas and methods, and specifically is CS, algorithms.
I'm a CS masters student. I'm taking a class in Computational Geometry. How much code has been presented to me in the class? None. Does this mean that I'm not getting a quality education? Hardly. The combination of pseudocode, description, theorems and proofs are far more useful for understanding a problem in whole than some single implementation would be. Implementations usually include details that are of little interest and would clutter the issue, obscuring the important bits.
Often, code created for research is not really production quality anyways, it is written specifically for a single problem. When code is high quality and sufficiently general purpose that it would actually be useful to others it often is released. In my field of computer graphics, code like this has been released which was used in writing several papers I have read and used.
While there is certainly some politics in CS publication, I doubt there is significantly more than any other academic field, and likely a good bit less than many "softer" fields.
You are correct that code is rarely published, that code quality may be poor, and that said code is probably not reusable. I say you're missing the point, the goal of CS is to create or discover new algorithms and methods, not write production quality code. If that's what you want I suggest you study software engineering (a branch of CS by most groupings, but with significantly different goals than most other areas). In any case, creating finished products is NOT the goal of academic research, and CS should not be an exception to this rule.
As far as not being verifiable, I say BS. The methods described in most CS articles easily give enough information for a person skilled and knowledgable in the field to duplicate the work. I've personally implemented a few fairly complex algorithms based only on what is described in a research paper. Additionally, duplicating a CS method takes considerably less effort than duplicating most experiments in the physical sciences. Everyone has access to computers, not many people have access to things like giant telescopes or particle accelerators, and the only resources consumed in algorithm implementation are time and electricity.
In conclusion, I'm not terribly worried about CS research. Lots of great projects are being done that fill both theoretical and practical voids. These ideas will filter through to practical applications and/or inspire more new ideas, and progress will march on.
Maybe that's not that much less than $100, but there's also no way you get by on $20. 2 Big Macs per day? That's what, $7 for the value meals and at least $4 just for the burgers? Already at about $30 per wekk or more.
In any case, I think that your health (not to mention the pleasure of eating real food instead of crap) is worh an extra few bucks a day. If not, then your budget likely needs some prioritizing, as you probably waste more money elsewhere (not because you specifically can't budget, but because almost everyone has some waste).
- He's new, so much of that week was probably also spent getting acclimated to the company, it's systems, and its people, which would have taken some time in any case.
- Maybe he's more productive in Linux. If that's what he uses best, then even if Linux only saved him one hour per week over OS X the Linux machine would pay back its setup costs in one year.
- Even if he would have eventually become more productive with a Mac, if he's not used to it there would be an initial learning time cost, even for a computer expert.
- Assuming an employee costs a company $100K/year (salary, benefits, utilities and other resources), one week would have covered about one powerbook.
I don't know what the best decision really would have been, but just like TCO analysis you can't get the big picture by only looking at one factor.Interesting. So you complain about how CC signatures are basically useless (no argument there), and then go on and bitch because some stores ask for a photo ID?
You do realize that a photo ID is much more effective in stopping CC abuse than a signature, don't you? As in most potential CC thieves don't look much like you, even if they could duplicate your signature (or get away with any old signature)?
Actually, what you're describing here is market research. Marketing is find ways to get people to buy a product that has already been created. Market research tends to benefit both producer and consumer by allowing creation of better products. Marketing has more one-sided benefits in favor of producers, as good marketing might be getting lots of people to buy an inferior or overpriced product.
I've used Mozilla for mail for the past several years, and I've had no problem trasnferring my POP folders to a new machine. I don't get all that much mail either, so the storage space wouldn't even be an issue, except...
Attachments that I no longer care about take up I'm sure the majority of the storage. I'd like to be able to delete the attachments while keeping the messages. Does anyone know of any tools that can go through mail in POP3 format and allow selective deleting of attachments. Ideally I'd like it to be a separate tool from an actual email program, but if there's an email program that can do this (and runs on Linux) I might consider setting it up on my existing folders just to allow deleting of attachments on occasion, while uisng Mozilla to actually download, send and read email.
Maybe your problem is that you only spent time with CS people.
I had a great time as an undergrad in CS. Did lots of partying (esp. living in the dorms freshman and sophomore years), played intramural flag football, and did well academically to boot. None of my friends were in CS though, mostly I met them through living in the dorms or something like that. This might be more difficult if you go to a completely tech oriented school, but at a more balanced school with large departments in the humanities and such it shouldn't be a big deal.
As long as your fairly self sufficient academically there's no real reason to heng out with people in your major any more than you want. Building contacts for future jobs maybe (although apparently even this didn't work out to well for you!), but that's about it.
I originally had you pegged for a troll from your initial snide remark, maybe I was wrong, we may actually see more eye to eye than I would have thought on this one. Still, I don't think it is a question of having it both ways. There are plently of people who feel corps have far to much power to control copyright who don't necessarily want to make unauthorized copies of stuff purely because they're too cheap to pay for it. If anything the corps are the ones who want to have it both ways, they want the benefits of government granted copyright protection without the trade off of eventual release into public domain, or enabling any semblance of fair use.
Identity vandalism is a somewhat better term than identity theft, but doesn't really convey the seriousness of the crime, and really there are elements of theft, vandalism, and outright fraud. It's like theft because their is only one identity, and it can be used by others for unauthorized purposes which benefit the identity thief (unlike pure vandalism). It is also like vandalism because real damage is caused by the time you get control of your identity back. It bears very little resemblance to copyright infringment (or piracy, or whatever term you want to use), though, which is what prompted my initial remark in the first place.
Nice try but no, jackass. Copyright infringement is not theft because me making a copy of some data does not alter the copy that you or anyone else has in their possession.
Identity theft is much closer to physical theft because their is only one record of a person's identity in any of these databases. When someone steals your identity it does not create a new copy in the database, it fucks up your one and only copy. Just like physical theft and abuse of an item fucks up the one one and only instance of the particular item that was stolen.
Maybe theft isn't the perfect term to describe it, but identity theft bears a lot more resemblance to physical property theft than unauthorized data copying does.
Yes, but the parent was stating a difference between apps which require 3D capabilities themselves vs. those that only require 3D like capabilities from the window manager.
2D apps treated like textured polys do not need their own depth buffer, frame buffer, etc. They just need a texture buffer, and the window manager treats them like texture polys in a single, comprehensive 3D app. Conventional 3D apps require their own depth buffer, frame buffer, textures, etc. in addition to that used by the window manager. I was pointing out that while the requirements for 30+ 2D apps are fairly large, they are considereably smaller than the same number of true 3D apps, because they have needs beyond the WM overhead.
All in all this is a pretty fair assessment, but I think it leans a bit to the pessemistic.
First, I doubt a window manager would actually use a 32 bit z-buffer. 8 bit would be overkill here (enought to specify a unique depth for 256 windows). Even a 3D window manager would get by on 16 bit depth no problem, I believe it's the most commonly used depth for most true 3D apps now.
Also, I doubt that in many cases more than a small number of true 3D windows would be needed. Someone who is working with 30+ windows open most likely has mostly terminals, web pages and text editors open, with maybe a few 3D apps.
Sorry, while I agree with the general thrust of this your 10 line rule seems overly pedantic and ultimately counterproductive.
Often I have written largeish functions such that while they could be broken into smaller functions, at the time there would be no purpose in doing so except to enforce the "only short functions allowed" rule. In other words the small functions would only be called from a single point in the code. In situations like this recursively breaking a function into smaller functions adds function call overhead which impacts performance, but more importantly adds additional whitespace and boilerplate function definition code to the source that can detract from readability. This is especially true if a developer has to mentally step through every line of code in a function: if the function is monolithic the step through is basically linear, while if the function is heavily subdivided the developer must jump back and forth in the code to follow the chain of execution.
Yes, yes, ideally each function should be tested and verified individually, but if a function was custom coded for a specific purpose only to be used in one larger function then this is often difficult to do, due to lack of alternative test situations.
It is often the case that a chunk of code that is only useful in one location early on may be useful in multiple places later, but I believe it is best to wait until this time to refactor the original code into smaller, more modular functions.
The flat line they are talking about is likely a rolling average. With a rolling average when the results are trully random the display will be flat. However when the results break away from randomness the average will tend away from the midline.
This would be the easiest way to graph deviations from true randomness.
The best reason is if you ever want to change the layout of your site in the future.
If done correctly and with foresight you can design complete a CSS layout which can be completely modified at a later date without a single change to your (X)HTML code. This isn't trivial, as it is hard to know for certain what changes you may want in the future.
The typicaly scheme which I use on my rather simply laid out website (monsterden.net) is basically to divide the code up into chunks using divs, and assign IDs or classes to divs based on content type. Doing this you will generally only run into problems if you decide at a later time that you want to rearrange blocks with finer granularity than you initially planned.
Another reason which does not apply in all cases is readability. If you look a the code on my site you'll see that it is extremely readable. This is important to me because I code my site completely by hand. CSS removes a lot of table clutter and makes nesting of code blocks less of an issue.
Finally you can do some rather nifty tricks with CSS that would otherwise require Javascript. On my background images pages I use a few of these to create some nice artistic effects, including the changing of the cneter image when a link is hovered over (NOTE: I'm not sure if the CSS on this particular part of my site works in MSIE, as I went for coolness over full compatability here. The rest of my site is fully compatible with all modern browsers AKAIK).