I read that article before. The fatal weakness of its reasoning is that it only focuses on fatalities. The reality is that even if you got ill with the flu, you almost never died (under 0.1% fatality rate). Even the super-fatal pandemic flu of 1918 was about 5% fatal among those sickened. I doubt if it is feasible to get a statistically significant count of fatalities in a controlled study sample.
But even if you do not die, flu is pretty costly. It is costly in the time you spend miserable, sick and out of action. It is costly to the colleagues, friends and family that you in turn sicken. It is costly to society as a whole. Vaccines either prevent that sickening altogether or reduce its severity. That makes vaccination campaigns valuable to society as a whole -- even to the unvaccinated -- because any flu case prevented or shortened will eliminate yet another infection source. Since flu spreads, well, virally, stopping even one source is significant. That's why govt agencies tend to be on board, because they are worried about the health of the overall society.
I know you hate Java, but do reconsider. Since you have established your.NET skills, you will probably cover the vast majority of the universe of software job opportunities if you can add Java to the list. Of course there is a lot valid stuff outside of.NET and Java, but those are the dominant players, professionally speaking.
You won't be locked into the Java language. The Java ecosystem is much larger than that: the Java class libraries, of course, but also Spring, Hibernate, J2EE, the Apache goodies, Eclipse, debuggers, profilers, monitoring etc. If you expect to integrate your software with a future unknown codebase -- perhaps by acquisition or merger -- there is a good chance it will be Java-based.
Ruby, Python, JavaScript and other languages do run on the Java JVM, but I suggest taking a look at languages that really sprung up from Java, like Groovy and Scala. Groovy appeals to the dynamic typing camp, and Scala to the static typing camp. The advantage with these languages is that they integrate into the Java platform just about as well as the Java language itself. You can write real Java-visible classes in these languages that integrate seamlessly with the rest of the Java platform, including annotations, generics and enums. Other languages can run in the JVM and call Java classes, but it's likely a one-way street. Java classes cannot call, say, a JavaScript class (JavaScript has no classes!).
My own experience has been with Groovy, which claims Smalltalk, Ruby and Python as its inspiration. It's much more productive, powerful and readable than Java. Yet it integrates well with all things Java. And yes, it's production code deployed on Linux.
For many suffering from medical depression, a key criteria for effective treatment with antidepressants is that -- duh -- you have to use stuff that works. Patients will try one drug after another before settling on something that works, for as long as that one works. Those that don't work? Besides being ineffective at treating the symptom, they also add a bunch of unpleasant side-effects. You won't exactly be feeling cheery using the wrong medication: you'd likely feel even more depressed.
That's the problem I have with this study: they don't address the fact that meaningful use of antidepressants only comes from selective use of the *right* antidepressants for each patient. Of course, selective sampling is exactly what statistical methods try to avoid. In fact, this study explicitly states that they include *unsuccessful* trials. Throwing all cases together like this may be statistically valid, but does not address the efficacy of *correct* use of antidepressants.
In short, this study does not enable us to come to conclusions that matter.
No, the problem is not one of acronym compliance. The pool of applicants is simply very shallow. It's hard to find *available* candidates who can demonstrate basic technical ability. People who interview for an engineering position with a 6 figure salary crash and burn when asked to reverse a linked list (it's not a critical question, but I've stopped asking that one). Very few seem to know their data structures and their performance characteristics. I'm talking about undergraduate data structures and algo stuff: arrays vs linked list, trees vs hash tables, big O notation. The heck with the acronyms.
It's not that there aren't good engineers. I meet lots at geek events. But they aren't looking. They are well compensated, or they have their own business, or are just plain satisfied. When we interview someone we like, they often accept another offer before we even bring them in for a second interview, let alone start waving money at them. We found the last two hires out of state.
So as the original reply said, please point me at that pool. It's a desert out here.
I sympathize with your situation, but I'm only reporting the situation as I see it. We have trouble finding good employees. Well-funded startups are aggressively hiring away the ones we do have. Recruiters have always been around, but they are working a lot harder now to round up candidates to place. Our positions remain unfilled. The candidates I've interviewed are still employed: the engineers we will hire will likely leave behind empty positions at their old places that need filling.
I cannot speak directly to your situation nor your geographical area, but there are all kinds of reasons why you may be having problems even when many jobs are available. You may not have the kind of experience currently in demand. Your network may truly be inadequate. Recruiters may be biased against the long-term unemployed (they favor people who are already working). You may be facing age discrimination.
I'm sure the tech job market is dysfunctional in some ways, if jobs remain unfilled and some people remain unemployed. I don't have easy solutions for those who fall between the cracks. But my point remains: there are jobs that need filling, and you won't find them in the papers nor job sites.
Publicize our jobs? We do -- on our corporate web site. It's not a question of what we would LIKE to do, but what we can AFFORD to do given available resources.
People who have tried the usual Monster/Dice/HotJobs sites know that you would get inundated with resumes, most of which are unsuitable. Resumebots are not very discriminating: the quality/relevance level is little better than spam. Wading through the mess of resumes, figuring out who are suitable and who are faking it, phone-screening the candidates... small companies simply don't have that kind of time and resources. We (engineers) spend hours talking to each candidate who comes in, taking valuable time away from our engineering work. If someone walks in that door, that someone better be worth our time. That is why we depend on word of mouth -- people that our own (good) employees can personally vouch for -- or from good recruiters who have already done the hard work of evaluating candidates.
This is not the ideal situation. But for companies our size, we simply don't have a better answer.
You're right. They are not in the want ads. They are not on the web sites. But there are plenty available, and we have mega trouble filling positions.
What you need to understand is that the market for experienced software engineers and the like is not wide open in the help wanted ads. People rely on word of mouth, personal references and recruiters. I can tell you that at my employer, we have exhausted our personal networks, even with a $2000 employee referral bonus constantly dangled in front of us. It's a recruiters game now. I've heard people talk about waving more money at recruiters to get better quality candidates. It's really that hard.
To give you more anecdotes: recruiters are aggressive. A few times, I overheard a recruiter call a neighboring colleague ("no thanks, I'm not really interested right now") and then hear my phone ring: it's the same recruiter. We've had a recruiter successfully bring a new hire here, then try to hire me away (she's been at me for months now). A former employee at a new, well-funded startup has hired away several of my valuable colleagues. Management are working hard to please and retain those still here. I went to a local Java Meetup, and found a recruiter mingling in looking for people to place. Mind you, these Meetups are small affairs with maybe a half dozen people, so that's slim pickings. The recruiter too shared the fact that good candidates are hard to find. The local Java user's group last month had several "we want warm bodies" announcements. I get inquiries out of the blue from my LinkedIn network looking for engineers.
For smaller companies like mine, recruiting experienced engineers is starting to look like recruiting executives. It's hard to find good, solid engineers who fit right technically and culturally in a fast-paced, exciting startup environment. Public job ads just get us lots of junk applications not much better than spam. So it's personal referrals and head hunters now.
Use your network. Impress your head hunter. If you are good and have the right skills, lots of people will climb over each other to find you.
Energy? It'll save plenty: this Vista box is the first desktop PC I have that effortlessly goes in and out of S3 standby mode. This is the suspend-to-RAM mode where almost everything else shuts down -- CPU, network, video, fans, HD -- and a tiny trickle goes into keeping the RAM alive. Yet it transitions in and out of standby mode in mere seconds. Vista makes it trivially easy and convenient to use a PC energy state that uses almost no energy, while it seems a bit of a black art with a Linux desktop. By bringing S3 to the masses, Vista has done a lot to save energy.
So the article is essentially an ad for the Flash platform as a substitute for JavaScript. I find it's excessively dismissive of JavaScript.
- Bruce assumes that we are reaching the limits of JavaScript. But are we? JavaScript has been around for ages. All it took was a catalyst, somebody (Google) to demonstrate to a wide audience that existing technology already in place could do so much more. Remember how you first felt playing with Google Maps? For me, it was "gee, I didn't realize you could do that with JavaScript". Heck, I never saw that kind of interface before even on a desktop app. But it only took a couple of guys a couple of hours to develop a clean implementation of Google Maps from scratch (Ajaxian Maps). JavaScript and browsers will continue to evolve, so the sky's the limit.
- Flash is not a substitute for Ajax. JavaScript controls the entire browser, subject to security restrictions. Flash is just an animation in a box. I repeat: in a box. You'd need a fundamental change in architecture for Flash to substitute for a browser's JavaScript engine. As it is, Flash is a pretty good *complement* to JavaScript. But don't mistake one for the other.
I have been using Linux for years, since Red Hat 5. At my last job, I architected and developed on a Linux-based platform complete with diskless touchscreen kiosks running off a single very modest server. I was also the Linux sysadmin, and years later I'm told that the DNS/SMTP/IMAP/WWW/Samba/NTP/SSH/NFS servers I set up for them are still running great on PCs so modest that Windows likely wouldn't boot on them. At home, I've tracked the latest releases of Red Hat, Mandrake and CentOS, playing with the latest KDE releases as they come out. So the general Linux platform is very solid and works fine. Alone.
The trouble is that everything I want to *DO* now has to be done in Windows.
- My parents abroad want to video conference. Getting GnomeMeeting/Ekiga to be friendly with Netmeeting has been a nightmare. With the last Gnomemeeting update, I've not been able to connect well for months. My parents have moved on to Skype, and it only does video on Windows. Moreover, the Quickcam webcam drivers for Linux (which I have to recompile for every kernel update) are horrible: lousy performance, bad color balance... I had no idea how bad they were until I tried installing Logitec's real drivers in a Win2K VMWare image.
- To get into my corporate VPN, I need to install a Windows driver. This is not negotiable.
- To access my office box remotely, I need RemoteAdmin for Windows.
- My wife (a music major) wants to purchase songs from the iTunes store. I've pointed her to eMusic (indie music on MP3s), but she says its too limited. No, iTunes is not feasible under Wine. Nor can iTunes find the iPod under VMWare Player.
- Everybody uses MS Office. Bosses, colleagues, partners, customers. I've tried to sing the praises of OpenOffice.org, but interoperability is limited. Documents never look right when transferring between OO and Office.
So while Linux is fine technically, my Linux PC is basically a useless box with a fan. To do anything useful, I need Windows. Whether it is my Win2K image running in a (slow) VMWare image or the win32 DLLs needed to make multimedia feasible or the Windows TrueType fonts that makes Linux look respectable or the WinXP box that I access remotely to do real work, I'm hopelessly tied to Windows. At some point came the realization that Linux offers me little value. I'm spending all my time in Windows anyway. An OS is supposed to boot up and get out of my way so I can get real work (or real fun) done. Instead, Linux is constantly holding me back. It simply does not play well with the part of the computing universe that matters to me. I'm not blaming anyone: it's just the reality of my situation.
So I'm now shopping for a new PC with Windows preinstalled. Farewell, Linux, you have been a faithful friend. Too bad I can't take you anywhere.
Does anyone know why Vista is such a resource hog? I don't mean the fancy UI/eye candy. I mean basic OS functionality: even Vista's most basic mode without the fancy features has a bare minimum RAM requirement of a half gig. At home, I have a Linux/KDE box with Windows 2000 running in a VMWare image -- hardly a minimal environment -- all with 384M of RAM. Apart from the exotic graphics stuff turned off, what is it about Vista that is hogging all that RAM? Can that junk be turned off?
Most of the time, I want an OS to boot up and get out of my way so I can open up my applications where I do my real work. I'm not sure I'm too excited about an OS that wants most of my RAM just to wake up, leaving me with little room to do real work.
The burn-to-CD-and-rerip workaround is not really a scalable workaround. For any significant music collection, you want to organize by the ubiquitous artist/album/track metadata. But you don't get that when you rip from a homebrew CD: all the CDDB tagging that we take for granted from commercial CDs won't be available. So you will have to enter them yourself: artist, album, track, for every single track that you rip. Either that, or live with a "Misc" folder of 672 files with informative names like "file00012848.mp3".
C's functions are totally context-free. You can't work with them as you would methods of related objects, where behavior is married to some presumed context. JS's functions are more OOPy, in that function calls are always associated with some object. A JS function can both retain context (through closures) and gain new context simultaneously when you reassign it to another object.
An illustration: you can change functions on an existing object, and functions receive the context in which they reside. So when you assign a function to an object:
var object = { x:1, y:2, sum:0, };// assign function here object.f = function(z) {
sum = this.x + this.y + z; }
that function sees the properties of the object to which it is assigned via "this".
Another difference is that you cannot readily change behavior in C by assigning functions. Because of the pointer/function distinction, you can only use function pointers with code that was written from the beginning to accept and execute functions via pointers. In JS, functions are a standard data type: all functions in all objects can be changed, and changing those functions will change the objects' behavior.
I wish people would stop using statistics exclusively from 2000, whether they be CS enrollments like here or related stats. In 2000, we were at the height of the tech bubble. Lots of people and money went into tech that should not have. In the case of people, that meant (among lots of other things... don't want to oversimplify) lots of CS majors who had no aptitude for CS. It's not a realistic number.
What I'd like to see are multi-year numbers that give us a better idea of the trend, both pre- and post-bubble. 2000 was an anomaly. 2000 was unsustainable. 2000 was when things went kablooey. We don't want to go there again in a hurry, so quit talking about it.
Too any job number comparisons are made to the last peak of 3-4 years ago. But that was the height of the dot.com bubble. That was the point where the industry was too crazy to sustain itself. Outrageous money was spent irresponsibly, qualified engineers were hard to find, classic measures of business sanity went out the window. We don't want to go there again. So if things don't look quite so rosy compared to those days, it's because we're not yet at the point of another dot-kablooie. So it's not like "the good old days"? Good.
What happened here is that a company had a product, fired the staff developing that product, and then still released a new version utilising the continuing free labour of those who it had put out of a job.
That's not a multiple choice question. There are winners and losers in many technological trends. The Luddites were right, in a way: they were losing their jobs, and someone else -- not them -- benefitted. It was a simple win-lose scenario, resolved in the case of the Luddites by mass hangings and other forms of repression.
There is no simple "solution" for the losers of any such trend. Innovation is usually the answer, except that it is a long term solution to a short term problem, meaning losers will continue to lose for a while. Career change is not easy: financial barriers exist where class barriers did earlier. Have you priced a college education lately? I guess the real answer is to grin and bear it.
Open source lowers the cost of doing business by substituting free labor for expensive developers.
Offshore outsourcing lowers the costs of doing business by substituting cheap labor for expensive developers.
Open source creates a few losers -- Microsoft, SCO -- but benefits many more by lowering costs and making stuff cheaper in general. "So let's screw the losers."
Offshore outsourcing creates a few losers -- some US workers -- but benefits many more by lowering costs and making stuff cheaper in general. "So let's screw the losers."
As one who rather likes open source, but whose job can potentially be offshored, I am having trouble making up my mind about this offshore outsourcing thing. I know there are other differences and complexities. The "free software" advocates want code to be free-as-in-speech, but the momentum is really behind the free-as-in-beer motive. Also, there are some who argue that offshore outsourcing will be detrimental to the US economy as a whole, but those who argue otherwise -- and back their arguments with data -- seem to have the better argument. So the above paragraphs distill the state of my reasoning at the moment. I have trouble seeing how I can favor one and oppose the other.
Please, argue with me.
(Pardon this repost... didn't get any response last time).
Open source lowers the cost of doing business by substituting free labor for expensive developers.
Offshore outsourcing lowers the costs of doing business by substituting cheap labor for expensive developers.
Open source creates a few losers -- Microsoft, SCO -- but benefits many more by lowering costs and making stuff cheaper in general. "So let's screw the losers."
Offshore outsourcing creates a few losers -- some US workers -- but benefits many more by lowering costs and making stuff cheaper in general. "So let's screw the losers."
As one who rather likes open source, but whose job can potentially be offshored, I am having trouble making up my mind about this offshore outsourcing thing. I know there are other differences and complexities. The "free software" advocates want code to be free-as-in-speech, but the momentum is really behind the free-as-in-beer motive. Also, there are some who argue that offshore outsourcing will be detrimental to the US economy as a whole, but those who argue otherwise -- and back their arguments with data -- seem to have the better argument. So the above paragraphs distill the state of my reasoning at the moment. I have trouble seeing how I can favor one and oppose the other.
The MTA's work is relatively light compared to what anti-spam software must do. This is especially true of SpamAssassin. While it does have some advantages over its competition, SpamAssassin is extremely resource intensive. Firstly, SpamAssassin is not written in fast C/C++ but Perl. Every email is sent through zillions of Perl regex rules. Then there is the Perl implementation of the Bayesian test, which really bogs down when an email auto-learned. Then there are the various (optional) network lookup tests: several RBLs, Pyzor/Razor/DCC... each email can eat up a lot of resources even if you bypass the startup overhead by running spamd.
I have also seen situations where SpamAssassin was not correctly respecting the maximum child spawn limit. Since spamd is a fairly heavyweight process, the server started swapping and throughput plunged.
Such heavy overhead is not a essential part of anti-spam software. Something NOT written in Perl nor any "interpreted" language, something with a smaller footprint, will be much, much faster. I wonder how many people have switched to dspam for this reason?
For any desktop user who might want to do videoconferencing, the lack of support for current cheap webcams (the Logitec Quickcams and no-names) is a major hindrance. I'm not downplaying the difficulties driver authors face: silently changing chipsets, proprietory compression formats... but who's at fault is not the issue. The problem is that consumer-grade webcams are mostly unusable unless you track down specific (possibly older) models based on their USB identifiers. Yuck. And even if you find something compatible, you are likely to see significantly inferior performance (fps, color balance, mic, snapshot button). I have been told to buy a webcam with its own IP address: basically a web-ready streaming video server that costs a couple of hundred dollars. That's a far cry from a $15 webcam from OfficeMax. Overall, the consumer grade Linux has a ways to go.
XFree86 has support for touchscreens: touch events get converted to mouse X events. The Elographics touchscreens work fine. In fact, Elographics has their own (enhanced) XFree86 drivers that you can request directly from them. Their calibration/diagnostics/setup utility is written in Java, so it works under Linux.
Slashdot Mirror
I read that article before. The fatal weakness of its reasoning is that it only focuses on fatalities. The reality is that even if you got ill with the flu, you almost never died (under 0.1% fatality rate). Even the super-fatal pandemic flu of 1918 was about 5% fatal among those sickened. I doubt if it is feasible to get a statistically significant count of fatalities in a controlled study sample.
But even if you do not die, flu is pretty costly. It is costly in the time you spend miserable, sick and out of action. It is costly to the colleagues, friends and family that you in turn sicken. It is costly to society as a whole. Vaccines either prevent that sickening altogether or reduce its severity. That makes vaccination campaigns valuable to society as a whole -- even to the unvaccinated -- because any flu case prevented or shortened will eliminate yet another infection source. Since flu spreads, well, virally, stopping even one source is significant. That's why govt agencies tend to be on board, because they are worried about the health of the overall society.
I know you hate Java, but do reconsider. Since you have established your .NET skills, you will probably cover the vast majority of the universe of software job opportunities if you can add Java to the list. Of course there is a lot valid stuff outside of .NET and Java, but those are the dominant players, professionally speaking.
You won't be locked into the Java language. The Java ecosystem is much larger than that: the Java class libraries, of course, but also Spring, Hibernate, J2EE, the Apache goodies, Eclipse, debuggers, profilers, monitoring etc. If you expect to integrate your software with a future unknown codebase -- perhaps by acquisition or merger -- there is a good chance it will be Java-based.
Ruby, Python, JavaScript and other languages do run on the Java JVM, but I suggest taking a look at languages that really sprung up from Java, like Groovy and Scala. Groovy appeals to the dynamic typing camp, and Scala to the static typing camp. The advantage with these languages is that they integrate into the Java platform just about as well as the Java language itself. You can write real Java-visible classes in these languages that integrate seamlessly with the rest of the Java platform, including annotations, generics and enums. Other languages can run in the JVM and call Java classes, but it's likely a one-way street. Java classes cannot call, say, a JavaScript class (JavaScript has no classes!).
My own experience has been with Groovy, which claims Smalltalk, Ruby and Python as its inspiration. It's much more productive, powerful and readable than Java. Yet it integrates well with all things Java. And yes, it's production code deployed on Linux.
For many suffering from medical depression, a key criteria for effective treatment with antidepressants is that -- duh -- you have to use stuff that works. Patients will try one drug after another before settling on something that works, for as long as that one works. Those that don't work? Besides being ineffective at treating the symptom, they also add a bunch of unpleasant side-effects. You won't exactly be feeling cheery using the wrong medication: you'd likely feel even more depressed.
That's the problem I have with this study: they don't address the fact that meaningful use of antidepressants only comes from selective use of the *right* antidepressants for each patient. Of course, selective sampling is exactly what statistical methods try to avoid. In fact, this study explicitly states that they include *unsuccessful* trials. Throwing all cases together like this may be statistically valid, but does not address the efficacy of *correct* use of antidepressants.
In short, this study does not enable us to come to conclusions that matter.
No, the problem is not one of acronym compliance. The pool of applicants is simply very shallow. It's hard to find *available* candidates who can demonstrate basic technical ability. People who interview for an engineering position with a 6 figure salary crash and burn when asked to reverse a linked list (it's not a critical question, but I've stopped asking that one). Very few seem to know their data structures and their performance characteristics. I'm talking about undergraduate data structures and algo stuff: arrays vs linked list, trees vs hash tables, big O notation. The heck with the acronyms.
It's not that there aren't good engineers. I meet lots at geek events. But they aren't looking. They are well compensated, or they have their own business, or are just plain satisfied. When we interview someone we like, they often accept another offer before we even bring them in for a second interview, let alone start waving money at them. We found the last two hires out of state.
So as the original reply said, please point me at that pool. It's a desert out here.
I sympathize with your situation, but I'm only reporting the situation as I see it. We have trouble finding good employees. Well-funded startups are aggressively hiring away the ones we do have. Recruiters have always been around, but they are working a lot harder now to round up candidates to place. Our positions remain unfilled. The candidates I've interviewed are still employed: the engineers we will hire will likely leave behind empty positions at their old places that need filling.
I cannot speak directly to your situation nor your geographical area, but there are all kinds of reasons why you may be having problems even when many jobs are available. You may not have the kind of experience currently in demand. Your network may truly be inadequate. Recruiters may be biased against the long-term unemployed (they favor people who are already working). You may be facing age discrimination.
I'm sure the tech job market is dysfunctional in some ways, if jobs remain unfilled and some people remain unemployed. I don't have easy solutions for those who fall between the cracks. But my point remains: there are jobs that need filling, and you won't find them in the papers nor job sites.
Publicize our jobs? We do -- on our corporate web site. It's not a question of what we would LIKE to do, but what we can AFFORD to do given available resources.
... small companies simply don't have that kind of time and resources. We (engineers) spend hours talking to each candidate who comes in, taking valuable time away from our engineering work. If someone walks in that door, that someone better be worth our time. That is why we depend on word of mouth -- people that our own (good) employees can personally vouch for -- or from good recruiters who have already done the hard work of evaluating candidates.
People who have tried the usual Monster/Dice/HotJobs sites know that you would get inundated with resumes, most of which are unsuitable. Resumebots are not very discriminating: the quality/relevance level is little better than spam. Wading through the mess of resumes, figuring out who are suitable and who are faking it, phone-screening the candidates
This is not the ideal situation. But for companies our size, we simply don't have a better answer.
"Where are these jobs?" [for experienced people]
You're right. They are not in the want ads. They are not on the web sites. But there are plenty available, and we have mega trouble filling positions.
What you need to understand is that the market for experienced software engineers and the like is not wide open in the help wanted ads. People rely on word of mouth, personal references and recruiters. I can tell you that at my employer, we have exhausted our personal networks, even with a $2000 employee referral bonus constantly dangled in front of us. It's a recruiters game now. I've heard people talk about waving more money at recruiters to get better quality candidates. It's really that hard.
To give you more anecdotes: recruiters are aggressive. A few times, I overheard a recruiter call a neighboring colleague ("no thanks, I'm not really interested right now") and then hear my phone ring: it's the same recruiter. We've had a recruiter successfully bring a new hire here, then try to hire me away (she's been at me for months now). A former employee at a new, well-funded startup has hired away several of my valuable colleagues. Management are working hard to please and retain those still here. I went to a local Java Meetup, and found a recruiter mingling in looking for people to place. Mind you, these Meetups are small affairs with maybe a half dozen people, so that's slim pickings. The recruiter too shared the fact that good candidates are hard to find. The local Java user's group last month had several "we want warm bodies" announcements. I get inquiries out of the blue from my LinkedIn network looking for engineers.
For smaller companies like mine, recruiting experienced engineers is starting to look like recruiting executives. It's hard to find good, solid engineers who fit right technically and culturally in a fast-paced, exciting startup environment. Public job ads just get us lots of junk applications not much better than spam. So it's personal referrals and head hunters now.
Use your network. Impress your head hunter. If you are good and have the right skills, lots of people will climb over each other to find you.
Energy? It'll save plenty: this Vista box is the first desktop PC I have that effortlessly goes in and out of S3 standby mode. This is the suspend-to-RAM mode where almost everything else shuts down -- CPU, network, video, fans, HD -- and a tiny trickle goes into keeping the RAM alive. Yet it transitions in and out of standby mode in mere seconds. Vista makes it trivially easy and convenient to use a PC energy state that uses almost no energy, while it seems a bit of a black art with a Linux desktop. By bringing S3 to the masses, Vista has done a lot to save energy.
So the article is essentially an ad for the Flash platform as a substitute for JavaScript. I find it's excessively dismissive of JavaScript.
- Bruce assumes that we are reaching the limits of JavaScript. But are we? JavaScript has been around for ages. All it took was a catalyst, somebody (Google) to demonstrate to a wide audience that existing technology already in place could do so much more. Remember how you first felt playing with Google Maps? For me, it was "gee, I didn't realize you could do that with JavaScript". Heck, I never saw that kind of interface before even on a desktop app. But it only took a couple of guys a couple of hours to develop a clean implementation of Google Maps from scratch (Ajaxian Maps). JavaScript and browsers will continue to evolve, so the sky's the limit.
- Flash is not a substitute for Ajax. JavaScript controls the entire browser, subject to security restrictions. Flash is just an animation in a box. I repeat: in a box. You'd need a fundamental change in architecture for Flash to substitute for a browser's JavaScript engine. As it is, Flash is a pretty good *complement* to JavaScript. But don't mistake one for the other.
I have been using Linux for years, since Red Hat 5. At my last job, I architected and developed on a Linux-based platform complete with diskless touchscreen kiosks running off a single very modest server. I was also the Linux sysadmin, and years later I'm told that the DNS/SMTP/IMAP/WWW/Samba/NTP/SSH/NFS servers I set up for them are still running great on PCs so modest that Windows likely wouldn't boot on them. At home, I've tracked the latest releases of Red Hat, Mandrake and CentOS, playing with the latest KDE releases as they come out. So the general Linux platform is very solid and works fine. Alone.
... I had no idea how bad they were until I tried installing Logitec's real drivers in a Win2K VMWare image.
The trouble is that everything I want to *DO* now has to be done in Windows.
- My parents abroad want to video conference. Getting GnomeMeeting/Ekiga to be friendly with Netmeeting has been a nightmare. With the last Gnomemeeting update, I've not been able to connect well for months. My parents have moved on to Skype, and it only does video on Windows. Moreover, the Quickcam webcam drivers for Linux (which I have to recompile for every kernel update) are horrible: lousy performance, bad color balance
- To get into my corporate VPN, I need to install a Windows driver. This is not negotiable.
- To access my office box remotely, I need RemoteAdmin for Windows.
- My wife (a music major) wants to purchase songs from the iTunes store. I've pointed her to eMusic (indie music on MP3s), but she says its too limited. No, iTunes is not feasible under Wine. Nor can iTunes find the iPod under VMWare Player.
- Everybody uses MS Office. Bosses, colleagues, partners, customers. I've tried to sing the praises of OpenOffice.org, but interoperability is limited. Documents never look right when transferring between OO and Office.
So while Linux is fine technically, my Linux PC is basically a useless box with a fan. To do anything useful, I need Windows. Whether it is my Win2K image running in a (slow) VMWare image or the win32 DLLs needed to make multimedia feasible or the Windows TrueType fonts that makes Linux look respectable or the WinXP box that I access remotely to do real work, I'm hopelessly tied to Windows. At some point came the realization that Linux offers me little value. I'm spending all my time in Windows anyway. An OS is supposed to boot up and get out of my way so I can get real work (or real fun) done. Instead, Linux is constantly holding me back. It simply does not play well with the part of the computing universe that matters to me. I'm not blaming anyone: it's just the reality of my situation.
So I'm now shopping for a new PC with Windows preinstalled. Farewell, Linux, you have been a faithful friend. Too bad I can't take you anywhere.
Does anyone know why Vista is such a resource hog? I don't mean the fancy UI/eye candy. I mean basic OS functionality: even Vista's most basic mode without the fancy features has a bare minimum RAM requirement of a half gig. At home, I have a Linux/KDE box with Windows 2000 running in a VMWare image -- hardly a minimal environment -- all with 384M of RAM. Apart from the exotic graphics stuff turned off, what is it about Vista that is hogging all that RAM? Can that junk be turned off?
Most of the time, I want an OS to boot up and get out of my way so I can open up my applications where I do my real work. I'm not sure I'm too excited about an OS that wants most of my RAM just to wake up, leaving me with little room to do real work.
The burn-to-CD-and-rerip workaround is not really a scalable workaround. For any significant music collection, you want to organize by the ubiquitous artist/album/track metadata. But you don't get that when you rip from a homebrew CD: all the CDDB tagging that we take for granted from commercial CDs won't be available. So you will have to enter them yourself: artist, album, track, for every single track that you rip. Either that, or live with a "Misc" folder of 672 files with informative names like "file00012848.mp3".
Ahem.
http://www.getfirebug.com/lite.html
Works with IE, Opera, Safari.
There is also a fairly powerful script debugger for IE that comes with Office.
Chris
C's functions are totally context-free. You can't work with them as you would methods of related objects, where behavior is married to some presumed context. JS's functions are more OOPy, in that function calls are always associated with some object. A JS function can both retain context (through closures) and gain new context simultaneously when you reassign it to another object.
// assign function here
An illustration: you can change functions on an existing object, and functions receive the context in which they reside. So when you assign a function to an object:
var object = { x:1, y:2, sum:0, };
object.f = function(z) {
sum = this.x + this.y + z;
}
that function sees the properties of the object to which it is assigned via "this".
Another difference is that you cannot readily change behavior in C by assigning functions. Because of the pointer/function distinction, you can only use function pointers with code that was written from the beginning to accept and execute functions via pointers. In JS, functions are a standard data type: all functions in all objects can be changed, and changing those functions will change the objects' behavior.
hris
I wish people would stop using statistics exclusively from 2000, whether they be CS enrollments like here or related stats. In 2000, we were at the height of the tech bubble. Lots of people and money went into tech that should not have. In the case of people, that meant (among lots of other things ... don't want to oversimplify) lots of CS majors who had no aptitude for CS. It's not a realistic number.
What I'd like to see are multi-year numbers that give us a better idea of the trend, both pre- and post-bubble. 2000 was an anomaly. 2000 was unsustainable. 2000 was when things went kablooey. We don't want to go there again in a hurry, so quit talking about it.
A work visa like a H1B is a lot harder and more hassle to obtain than a non-work visa like a tourist visa.
Too any job number comparisons are made to the last peak of 3-4 years ago. But that was the height of the dot.com bubble. That was the point where the industry was too crazy to sustain itself. Outrageous money was spent irresponsibly, qualified engineers were hard to find, classic measures of business sanity went out the window. We don't want to go there again. So if things don't look quite so rosy compared to those days, it's because we're not yet at the point of another dot-kablooie. So it's not like "the good old days"? Good.
It's called outsourcing.
Everybody's right. The issue is that costs and benefits are distributed unevenly.
"[XYZ] will eliminate jobs."
Answer: "[XYZ] will eliminate your jobs, but will drastically reduce our costs and benefit society as a whole."
[XYZ] is:
That's not a multiple choice question. There are winners and losers in many technological trends. The Luddites were right, in a way: they were losing their jobs, and someone else -- not them -- benefitted. It was a simple win-lose scenario, resolved in the case of the Luddites by mass hangings and other forms of repression.
There is no simple "solution" for the losers of any such trend. Innovation is usually the answer, except that it is a long term solution to a short term problem, meaning losers will continue to lose for a while. Career change is not easy: financial barriers exist where class barriers did earlier. Have you priced a college education lately? I guess the real answer is to grin and bear it.
Or saying "offshore outsourcing is causing unemployment for people who wish to do the same work for more pay".
As one who rather likes open source, but whose job can potentially be offshored, I am having trouble making up my mind about this offshore outsourcing thing. I know there are other differences and complexities. The "free software" advocates want code to be free-as-in-speech, but the momentum is really behind the free-as-in-beer motive. Also, there are some who argue that offshore outsourcing will be detrimental to the US economy as a whole, but those who argue otherwise -- and back their arguments with data -- seem to have the better argument. So the above paragraphs distill the state of my reasoning at the moment. I have trouble seeing how I can favor one and oppose the other.
Please, argue with me.
(Pardon this repost ... didn't get any response last time).
As one who rather likes open source, but whose job can potentially be offshored, I am having trouble making up my mind about this offshore outsourcing thing. I know there are other differences and complexities. The "free software" advocates want code to be free-as-in-speech, but the momentum is really behind the free-as-in-beer motive. Also, there are some who argue that offshore outsourcing will be detrimental to the US economy as a whole, but those who argue otherwise -- and back their arguments with data -- seem to have the better argument. So the above paragraphs distill the state of my reasoning at the moment. I have trouble seeing how I can favor one and oppose the other.
Please, argue with me.
The MTA's work is relatively light compared to what anti-spam software must do. This is especially true of SpamAssassin. While it does have some advantages over its competition, SpamAssassin is extremely resource intensive. Firstly, SpamAssassin is not written in fast C/C++ but Perl. Every email is sent through zillions of Perl regex rules. Then there is the Perl implementation of the Bayesian test, which really bogs down when an email auto-learned. Then there are the various (optional) network lookup tests: several RBLs, Pyzor/Razor/DCC ... each email can eat up a lot of resources even if you bypass the startup overhead by running spamd.
I have also seen situations where SpamAssassin was not correctly respecting the maximum child spawn limit. Since spamd is a fairly heavyweight process, the server started swapping and throughput plunged.
Such heavy overhead is not a essential part of anti-spam software. Something NOT written in Perl nor any "interpreted" language, something with a smaller footprint, will be much, much faster. I wonder how many people have switched to dspam for this reason?
For any desktop user who might want to do videoconferencing, the lack of support for current cheap webcams (the Logitec Quickcams and no-names) is a major hindrance. I'm not downplaying the difficulties driver authors face: silently changing chipsets, proprietory compression formats ... but who's at fault is not the issue. The problem is that consumer-grade webcams are mostly unusable unless you track down specific (possibly older) models based on their USB identifiers. Yuck. And even if you find something compatible, you are likely to see significantly inferior performance (fps, color balance, mic, snapshot button). I have been told to buy a webcam with its own IP address: basically a web-ready streaming video server that costs a couple of hundred dollars. That's a far cry from a $15 webcam from OfficeMax. Overall, the consumer grade Linux has a ways to go.
XFree86 has support for touchscreens: touch events get converted to mouse X events. The Elographics touchscreens work fine. In fact, Elographics has their own (enhanced) XFree86 drivers that you can request directly from them. Their calibration/diagnostics/setup utility is written in Java, so it works under Linux.