Those of you that seem to support the president have no problem buying our forced "charity" to the Iraqi people, but when it comes to health care for children, you all scream like someone is stealing your money.
I think Bush's economic principals have obviously failed considering the record high unemployment rates.
Clinton had the economy thing down. He didn't have two years of record high unemployment after enacting his economic plan.
You really don't have a any evidence to support that either are bad ideas. The US is likely to soon be one of a handful of countries without a universal health care system. It all goes back to big business buddies of the Republican party. If the government did universal health care, they would likely also be concerned about getting drugs for a normal price instead of the Republican backed drug monopoly prices that we see today.
Zucinich and Braun are my favorite candidates this coming election, but I would settle for Dean or Edwards.
I think you just need to watch it again to catch the philisophy, like what is actually the meaning of love? Why can programs feel it?
As for Trinity dieing, there is no other way he would do what he had to do at the end knowing she was still out there. The machine's didn't bring him back to life. Just think about what the oracle says about Smith's relation to Neo and the first movie seen where he infects Smith.
I think most of the people that watched the show and gave it a bad review just haven't had time to put the three movies together. I personally can't wait to get some DVD's and watch them all together to help catch the missing details instead of complaining that scenes were not relavent. Just like the first movie, I think I've experienced some kind of mental overload from all thats going on. I don't really get the movie yet, but I think having a Matrix day would probably clear it up. I'ld rather have a hard time understanding a movie (like kubric films) than have the film be entirely superficial (like any of the new Star Wars movies).
Anyhow, I expected I would have to see it a few times before I put the entire story together, and it always helps to watch it again from the very beginning. Anyone remember the copper top joke from Matrix 1, or the confusion from the phone booth scene?
Well, I do get a little afraid of XDoclet and the likes as well. You can blame most of it on the J2EE spec. You have to create 3 files and modify at least one more in about 3 different places in order to make a single bean. If the spec was more self contained it wouldn't have lead to XDoclet, which is now being used for more than it should be. XDoclet is for those people that don't believe in an IDE, or don't believe that IntelliJ IDEA is worth the price. What we've always needed is just a decent IDE or a more compact spec. for J2EE. I'm sure it will get better eventually, because XDoclet is just a cludge trying to cover the complexity issue of J2EE.
I can tell you from experience that Java's API is much more robust than perls for just about anything (even regular expressions are faster than perl). Java has JSP, taglibs, Struts, and Webwork whereas Perl has mod perl, and templates. All of the web scripting examples I gave with Java work well in a MVC architecture (which Perl and PHP only have templates for). So, from a maintenance and development perspective Java is pretty much superior (I know because I've used Perl, PHP, and JSP/Servlets). From a robustness perspective, for anything more than a database driven site, Java has been far more reliable and secure (I can't count the times that despite proper configuration of that perl form mail script, it has been used to spam. JavaMail API supports mime attachments, pop3 and imap with the same API, etc.). From a speed perspective, with the proper JVM (try JRocket, IBM, and SUN's because each web server will preform differently with each) and the proper servlet container (Jetty for free, or Resin for pay), the right database server/JDBC driver (you don't have a lot of options for the driver usually, but they matter more than you can imagine), you can out-do PHP and Perl without a doubt. From a scalability/reliability perspective, you can replicate session variables over a cluster (without hitting the DB, which means very much more scalable on a cluster) which means if a node on a properly configured Java system goes down, the end user will never even know, so I really can't think of a single other system that does this other than Java.
So, I, a developer, can tell you, without a doubt, that using perl or PHP for any application other than your home page is generally a bad idea. It's slower, less maintainable, doesn't scale as well, and is less secure than Java Servlets/JSP on average. You'll have a hard time fighting with numbers or playing with definitions in order to dispute that too. Believe me, because I've used PHP on Enterprise level projects, but only because it was better than ASP, and before I learned Servlets/JSP. I'm in the process right now of converting an application from Perl/C/C++/PHP to Java because it's is a mess to administer. C/C++ calls to the Win32 API are a nightmare because the documentation blatantly lies, and without having a windows client, we would be screwed from the start. Perl and it's DBI mess crashes so much that I had to write a bash script to keep it up (several versions of both the DBI and Perl and MySQL still will crash when you try to keep it up 24/7). PHP, being the best of the three, is being thrown away because once I actually had to learn Java to replace the others, I found that it's just superior to PHP for the kind of work I need it for (user editable web scripting).
I've researched it all from Delphi, C++Builder, VC++, VB, PHP, Perl, Lisp (and frineds), and Python before I started the project. I can tell you the problems you'll run into in each of them. Some of them absolutely have no purpose (VB, Delphi, C++Builder) but others are only suited for other tasks that don't grow beyond a simple program/application into a system of interconnected peices. You can't write a true 3 tier system in PHP, and there are good reasons you would want to. Java technologies include J2EE (a superclass of servlets) that gives you an enterprise application server that you don't have to write yourself. Python is the most promising with Zope, and I keep my eyes on it all the time, but just don't be so closed minded to think that Perl can do anything. It's not meant to, and there just aren't enough good and stable API's to work on a project so large that you need everything from Message Queueing and MIME Email to Serial Port IO and TAPI. Perl doesn't even have a Zope like entity. When Python gets Zope down right and starts expanding it's API's to support all the features I need, then I'll be reconsidering.
You're career could end quick if you fail to let the mentality that Perl/PHP is good enough for any job die. It's just not the true. You may be like
I hate always being the one to laugh at things like this.
Java saves the user money at the cost of the developers being ridiculed by people on slashdot that praise PHP (which is generally slower) than Java Servlets (the topic of conversation).
Maybe you think websites should be written in C and linked directly to the kernel? That would be a hard arguement for the likes of you to argue.
Re:Worth Learning?
on
Bitter EJB
·
· Score: 4, Insightful
I'm going to attempt to answer this question, but it's not really easy.
The first thing you should think about is pre-J2EE history. Back before J2EE everyone was developing their own systems. They had their own persistance layer, their own distributed transaction system, their own network protocols in some situations. Was this system faster? Yes, of course it was. If you can develop your own system custom to your task, it's always going to be faster.
Performance was not the reason for J2EE. Scalability was not the reason for J2EE. They are part of a new push in the software world. Overall economics is where J2EE comes in. It's more expensive to create your own framework than to buy one. It's more expensive to train developers on how to use your framework than an existing standard. If you had one person developing just your infrastructure full time it would cost you 100K/y in salary, which is more expensive than the most expensive J2EE server will run you on average.
Now I would like you to think, "Why don't we write our own database systems?" If we write our own system we wouldn't need to parse SQL. We could store the data in relation formats where it's good, and hierachial where it makes more sense to be hierarchial. The system would be faster be a very significant amount.
The point is if you keep "reinventing the wheel" (I hate that phrase.) your competitors will beat you to market.
I think I've covered two of the main points here. 1) It can be cheaper 2) It can be quicker to market.
Who cares if it requires a cluster of Opterons to run, software has always cost more than mass produced silicon waffers.
So, back to the main answer, which by now you probably can see, when to use EJB. 1) If you need distributed transactions. 2) If you need platform independance. (.Net clause) 3) If it will let you get to market quicker. 4) If you need declaritive security checks.
If none of those ifs apply to your project, then you would be an idiot riding hype if you use EJB. As for my EJB projects, I can generally say I need all 4. You can't force Linux on clients because they don't have administrative people that will patch it like they do windows. You can't use windows because it isn't reliable as a server generally. Let them pick and live with their own decision is my philosophy.
Still won't work... I would "bug" the lights near where emergency vehicles pass frequently and gather the info using IR as I drove by. Probably a 5$ disposable item, and I would collect the days key the first time it was used. You'ld have to put cameras and cops at lights to figure it out, then again, I might be able to do it from further away than just the intersection as well. I may even be able to make some buddies with the right people and set up the janitor with the system if he lets me make it. Also keep in mind that the bug I'm talking about would be about the size of dime as well. It would only be an ATMEL (one of the 8-pin chips), a LED, a photo transistor, and a coin cell. If it only transmits IR once a day, it would probably last at least 3 months, probably 6-12.
The best way to do it is the challenge method and RF, not IR. This would help track the progress of emergency vehicles as well and could be wired directly back to 911 so they can tell how far away the vehicle is. If you put one on your car, 911 tracking systems would see it, then you would get busted pretty fast. Nothing beats wiring the whole system up properly with security in mind to begin with. Eventually you would request a per incident key through hardwired land lines that are transmitted over using site keys that are carried out to the site so that loosing a key anywhere is a simple matter to fix.
It's not capillary action really... it is cohesion to the ball. The ball is rotated, and the ink would rather be on paper than the ball, so the ink leaves, thus creating a void which sucks in more ink.
There are some real space pens that don't use nitrogen pressure and can be used in both freezing cold and blistering heat because the ink is actually a near solid. The ball's friction tears off the ink as it goes, and the suction (the ball is mainly used as a valve in almost all ink pens) pulls ink in without letting air into the ink well.
There is a folk tale about ink pens and the cold war that I can't verify. It had to do with the US spending 5 million dollars or more in research to develop a pen that was cheap and would work in space while the Russians just used pencils. If anyone can find the origin of the folk tale, or actually find evidence for this story, I'ld love to read more.
I wonder what effects things like this has on the jewelry industry. In the next century, there will probably be children's toys that make and cut diamonds into any shape they can imagine with their 3d holographic editing system:)
Yeah, I'm interested in their magical switches too, because the network is always the bottleneck on a cluster. It's good we have a few people researching though.
I was just pointing out that if it's 40X the performance, and 130X the price, that's not two unheardof. Hopefully someone will do an Opteron cluster.
I just see similar comparisons in J2EE performance as well on clusters of machines. Someone will come out with the best price per performance offor for enterprise machines that doesn't scale beyone 2 machines and they claim that it is because the application server is superior.
BTW, I really love that you guys used Povray to test the cluster, even though any rendering that takes 2.5s is probably all network latency:). I have some Pov Files I created that take me 8h to render on my little 3 machine cluster. You should try cheating to render a movie perhaps. When I was at EKU (this was before both the clusters at UK) we put together 15 486's as a test to see what kind of computer could be built with retro hardware from the trash can. We ended up having the a decent, modern "computer" for free.
I think clusters would be a lot easier if Linux supported equal cost multipath routing easily and effeciently. Using PVM to route requests is a little crazy. I wrote my own network communication most of the time.
Oh, another thing that probably helps the Big Mac is their support of a faster PCI bus (generally macs support both 64bit and 66mhz) which can limit some Gbit cards.
32bit*33mhz = 1Gbit to share between all the network cards (for single PCI bus systems). 4X the bandwidth I'm sure would help.
Yes, but you make the common mistake that we are working on a linear scale here. Back when they were testing a portion of Big Mac they were getting twice the effecience they are now. If you go look at the exact flops and divide the cluster in costs, you'll see that you are getting very close to the KASY0's performance/price ratio.
Technically, the highest performance/price ratio would probably be a cheap chinese/tiwanese watch, but those APEX DVD players probably aren't far behind. I guess if you really measured it all out, the best performance/price ratio is probably an XBox since MS is loosing money on them.
When the KASY0 is doing TFLOPS, then maybe you can compare the two.
(Not trying to knock UK, I have many friends that worked on that cluster... I live about 15 minutes away.)
Yeah, but it shouldn't still be classified, and that's the point. There are investigators working for Sci-Fi channel that make up scenario's every day, and they can't figure out why it would still be classified. At least I think we deserve some kind of reason why reports on a 1965 incident are still classified.
The only other plausible theory I've heard is that it might contain information that would damage foreign relations. Still they could release a report saying what the hell it was without giving away it's contents. They've done nothing thus far, so it's degraded into an issue only resolvable by a judge.
They've actually researched enough and have some first hand accounts that it was not a U-2 crash. Even one person here on/. lived close to the incident and spoke to a firefighter that went out there before the others arrived, and described it as an acorn shaped object. They've elimenated the possiblity that it was a Satellite by talking to NASA themselves (RTA). You're theory is very unlikely. If you know anything about the Kecksburg incident, you'll know that civilians where held at gun point, and one reporter ended up dead. So, I think we have a right to know what the hell happened there at least to investigate the murder of a reporter.
I just don't like the fact that/.'ers come post here so skeptical that they hate skeptics without even knowing their reasons. I hate most when an idiot laughes at UFO researchers saying they are loons without even seeing the evidence they have.
Think of it this way. If an alien landed today, and shook hands with the locals, would you believe the media? If not, you are severely biased the other way. Either admit that you don't know what happened, which means investigating is a good thing, or admit that you are a horribly biased bastard and don't post. Posting extreme skepticism is not helping any debate.
(This isn't directed completely at you blair1q, I just summed up my problem with about half of the +4 insightful posts here.)
This exact post is in every PHP/JAVA story on/., all the links are usually broken, and the writer is an idiot.
Java was written in C as well. Java compiles directly to machine byte-code on load (which happens once per class, which means only once per jsp file just like PHP).
My real question is how does this troll keep getting modded up?
Just for your information, I had to switch from PHP to Java because PHP was impossible to maintain.
Actually, that's his free speach. I could tell everyone where to buy drugs, and that would be perfectly legal. If you have a problem with copyright infringement, take it up with those infringing, besides, now you know where to find them.
I agree with you, developers generally don't talk to the customer. If you had actually talked to several customers (I have to deal with nearly 30 in my main project), I think you would get the point that the customer generally doesn't know what they want. After years of dealing with them, I've gotten a handle on heading off their requests with features before they can ask for them usually. It saves time in the long run if you design a flexible model to begin with; instead of scrapping portions of the project, adapt them. Make your model vague. My favorite is creating N:M relationships with the relation table containing an ID defining the relationship. That way you can relate different sets of objects in different ways, but the same code can show which objects are related and how.
Basically, I like to believe that XP is a few notches too far up on the design issue, but not because the customer knows what they want, but rather because you can design a flexible archetecture. I've only met one customer that knew what they wanted, and that was because they wanted a duplicate of something else, except they wanted it stable and fast.
yeah, but most of OpenGL is mapping textures stored on the card to pixels. If anything, this shows how a system built with this in mind (the root of the story) could indeed be much faster than it being a side-effect. The main point is that X has very, very little to do with OpenGL rendering that a game like Quake does.
The more likely scenario was that the remote machine was rendering the OpenGL locally, then sending it back to the client. The network would support up to 20fps at 640x480x16. If it had just loaded the textures onto the card on the local machine, then transferred only the Vectors, I'm sure we can both agree that 100fps would not be unreasonable for even a remote machine on a 100mbit network because there is no concievable way that it takes 100mbit* 1s/100fps=1mbit/f=976K per frame to transfer vertices and instructions. If you look at the benchmarks for cards, you'll see very quickly that bus speed to the card doesn't do much except let the game load textures onto the card quicker.
In fact, if someone wants to argue enough about this, I may be driven to enough rage to write a stub opengl.so to put in your quake3 directory that just sends the function calls through some home brew RPC protocol to a client X app on the other end. Hell, I bet I could use XML and still see full frame rates.
I never multitask during games by choice. Some applications demand attention. I've moved every one of those applications onto my cheap laptop. Let's me run Gaim (talk to people while I camp), Mozilla (for cheat codes of course), Mozilla Mail (in case something important actually does come up), and lots of other business-esque software so I can see who's calling (from the phone system) to see if I need to mute before I pick up.
All the tweaking gamers out there should love this CD thing, for now. Then at some point they'll say, "Hey, you, Linux Geek! Can I copy this to my hard drive?" and of course you install a very compact Linux distro so that nothing else eats the precious CPU. Next you set up Mozilla for them so they don't have to reboot to browse the web. You set up Evolution, then they can get their mail. You set up StarOffice, and they can do their homework. Gamers won't want to reboot to windows except for the games that require it eventually (because they'll be slightly slower... at least that is true with some games like Unreal 1, and anything Quake). More demand for games on Linux (and therefore optimized drivers) is what it's all about, and I'm all for it. Now if NeverWinter would just conform to my theory that Linux is faster than Windows, I'ld be a very happy person!:)
Hey, I know some people, maybe I try to get Anand (of Anandtech) to run Linux benchmarks on cards to see who has the best Linux drivers. Someone did one back in 2000, but that's a lot dated. If it was in the benchmarks, I think it would help. (I'm very pleased with the NVidia drivers so far, but I haven't compared them to windows as much as I should. I think they use pretty much the same code because OpenGL is OpenGL even on Linux.)
Do you really want geeks showing up on your doorstep in full chainmail, a sword, and the amulet of Yendor? I think you should very much encourage them to keep their fantasies stored as harmless electronic and magnetic charges far, far away from this "Real Life". I mean, have you ever actually been to a LARP (Live-Action Role Playing)? It's definately not pretty. I know a guy who ended up getting stitches in a hospital. I guess he thought he had more hit points than he actually did. That's ok, he eventually did end up back at town center, and all his digits were intact and functional after seeing the healer. I think he gave up adventuring though, and he now works answering other adventurers' questions about these magical devices that help in the creation of scrolls at Lexmark.
Now, do you really want us role playing geeks to invade real life? You know these people that role play are usually the ones that right your software. Next time you check your X11 log for the last error to see if it's complaining about the mouse or display modes and all you see is runes, you'll learn to appreciate the fact there is a time and place for these fantasies. I personally don't want to have to describe my problem on the mailing lists using Quenya.
I guess in your development model you write more than you read?
I bet you even are going to say that you use more transactions than you query.
Finding/Reading/Loading data is always going to be the key to the best performance. Transactions are second to that. 3000X faster at reading affords it a lot of breathing space when it comes to transactions because everyone visiting slashdot is reading, but only a few post. This is the real world model.
I'm skeptical when it comes to easily retrieving and indexing data for reporting. You need to be able to query aggregate (yes, this is easy enough to code) but you need to join related objects. Do I have to do my own query planning/joining? What's the transaction API like? Do I lock objects then update them? What about deadlock? (Easy enough to avoid by anyone smart enough to know any of the algorithms like having to sort the order in which you lock resources, but still a hassle.) So many questions!
I didn't want to do it yet, but I think I'm going to start running j2ssh. I don't know how stable it is at this point, does anyone know? I'll run it on different port and test it for a while I guess.
You can get the LGPL SSH2 server/client written in Java from http://3sp.com/products/sshtools/sshtools.php
Take a laser and move it side to side. Shine this on a mirror that is tilted not on the plane that the light is traveling and you'll see that it isn't side to side anymore.
Dishes weren't what I was talking about refracting the RF. Objects around the antenna like trees, cars, and houses seem to do a great job is dispersing signals. A lot of our signal before we used a dish was being bounced off of houses across the street then back to the target.
A dish may only be reflecting signal but it gets mostly signal directly from the point of origin because it reflects RF away that isn't striking the dish almost perpendicular to the plane that runs tangent to it's vertex.
It makes sense to me now that I realize thought that the 99% bulk of the signal a dish picks up is not bouncing off of houses and cars and the Earth or ionosphere. This is generally the case with every other antenna including Yagi style.
Slashdot Mirror
Those of you that seem to support the president have no problem buying our forced "charity" to the Iraqi people, but when it comes to health care for children, you all scream like someone is stealing your money.
I think Bush's economic principals have obviously failed considering the record high unemployment rates.
Clinton had the economy thing down. He didn't have two years of record high unemployment after enacting his economic plan.
You really don't have a any evidence to support that either are bad ideas. The US is likely to soon be one of a handful of countries without a universal health care system. It all goes back to big business buddies of the Republican party. If the government did universal health care, they would likely also be concerned about getting drugs for a normal price instead of the Republican backed drug monopoly prices that we see today.
Zucinich and Braun are my favorite candidates this coming election, but I would settle for Dean or Edwards.
I think you just need to watch it again to catch the philisophy, like what is actually the meaning of love? Why can programs feel it?
As for Trinity dieing, there is no other way he would do what he had to do at the end knowing she was still out there. The machine's didn't bring him back to life. Just think about what the oracle says about Smith's relation to Neo and the first movie seen where he infects Smith.
I think most of the people that watched the show and gave it a bad review just haven't had time to put the three movies together. I personally can't wait to get some DVD's and watch them all together to help catch the missing details instead of complaining that scenes were not relavent. Just like the first movie, I think I've experienced some kind of mental overload from all thats going on. I don't really get the movie yet, but I think having a Matrix day would probably clear it up. I'ld rather have a hard time understanding a movie (like kubric films) than have the film be entirely superficial (like any of the new Star Wars movies).
Anyhow, I expected I would have to see it a few times before I put the entire story together, and it always helps to watch it again from the very beginning. Anyone remember the copper top joke from Matrix 1, or the confusion from the phone booth scene?
Well, I do get a little afraid of XDoclet and the likes as well. You can blame most of it on the J2EE spec. You have to create 3 files and modify at least one more in about 3 different places in order to make a single bean. If the spec was more self contained it wouldn't have lead to XDoclet, which is now being used for more than it should be. XDoclet is for those people that don't believe in an IDE, or don't believe that IntelliJ IDEA is worth the price. What we've always needed is just a decent IDE or a more compact spec. for J2EE. I'm sure it will get better eventually, because XDoclet is just a cludge trying to cover the complexity issue of J2EE.
I can tell you from experience that Java's API is much more robust than perls for just about anything (even regular expressions are faster than perl). Java has JSP, taglibs, Struts, and Webwork whereas Perl has mod perl, and templates. All of the web scripting examples I gave with Java work well in a MVC architecture (which Perl and PHP only have templates for). So, from a maintenance and development perspective Java is pretty much superior (I know because I've used Perl, PHP, and JSP/Servlets). From a robustness perspective, for anything more than a database driven site, Java has been far more reliable and secure (I can't count the times that despite proper configuration of that perl form mail script, it has been used to spam. JavaMail API supports mime attachments, pop3 and imap with the same API, etc.). From a speed perspective, with the proper JVM (try JRocket, IBM, and SUN's because each web server will preform differently with each) and the proper servlet container (Jetty for free, or Resin for pay), the right database server/JDBC driver (you don't have a lot of options for the driver usually, but they matter more than you can imagine), you can out-do PHP and Perl without a doubt. From a scalability/reliability perspective, you can replicate session variables over a cluster (without hitting the DB, which means very much more scalable on a cluster) which means if a node on a properly configured Java system goes down, the end user will never even know, so I really can't think of a single other system that does this other than Java.
So, I, a developer, can tell you, without a doubt, that using perl or PHP for any application other than your home page is generally a bad idea. It's slower, less maintainable, doesn't scale as well, and is less secure than Java Servlets/JSP on average. You'll have a hard time fighting with numbers or playing with definitions in order to dispute that too. Believe me, because I've used PHP on Enterprise level projects, but only because it was better than ASP, and before I learned Servlets/JSP. I'm in the process right now of converting an application from Perl/C/C++/PHP to Java because it's is a mess to administer. C/C++ calls to the Win32 API are a nightmare because the documentation blatantly lies, and without having a windows client, we would be screwed from the start. Perl and it's DBI mess crashes so much that I had to write a bash script to keep it up (several versions of both the DBI and Perl and MySQL still will crash when you try to keep it up 24/7). PHP, being the best of the three, is being thrown away because once I actually had to learn Java to replace the others, I found that it's just superior to PHP for the kind of work I need it for (user editable web scripting).
I've researched it all from Delphi, C++Builder, VC++, VB, PHP, Perl, Lisp (and frineds), and Python before I started the project. I can tell you the problems you'll run into in each of them. Some of them absolutely have no purpose (VB, Delphi, C++Builder) but others are only suited for other tasks that don't grow beyond a simple program/application into a system of interconnected peices. You can't write a true 3 tier system in PHP, and there are good reasons you would want to. Java technologies include J2EE (a superclass of servlets) that gives you an enterprise application server that you don't have to write yourself. Python is the most promising with Zope, and I keep my eyes on it all the time, but just don't be so closed minded to think that Perl can do anything. It's not meant to, and there just aren't enough good and stable API's to work on a project so large that you need everything from Message Queueing and MIME Email to Serial Port IO and TAPI. Perl doesn't even have a Zope like entity. When Python gets Zope down right and starts expanding it's API's to support all the features I need, then I'll be reconsidering.
You're career could end quick if you fail to let the mentality that Perl/PHP is good enough for any job die. It's just not the true. You may be like
I hate always being the one to laugh at things like this.
Java saves the user money at the cost of the developers being ridiculed by people on slashdot that praise PHP (which is generally slower) than Java Servlets (the topic of conversation).
Maybe you think websites should be written in C and linked directly to the kernel? That would be a hard arguement for the likes of you to argue.
I'm going to attempt to answer this question, but it's not really easy.
The first thing you should think about is pre-J2EE history. Back before J2EE everyone was developing their own systems. They had their own persistance layer, their own distributed transaction system, their own network protocols in some situations. Was this system faster? Yes, of course it was. If you can develop your own system custom to your task, it's always going to be faster.
Performance was not the reason for J2EE. Scalability was not the reason for J2EE. They are part of a new push in the software world. Overall economics is where J2EE comes in. It's more expensive to create your own framework than to buy one. It's more expensive to train developers on how to use your framework than an existing standard. If you had one person developing just your infrastructure full time it would cost you 100K/y in salary, which is more expensive than the most expensive J2EE server will run you on average.
Now I would like you to think, "Why don't we write our own database systems?" If we write our own system we wouldn't need to parse SQL. We could store the data in relation formats where it's good, and hierachial where it makes more sense to be hierarchial. The system would be faster be a very significant amount.
The point is if you keep "reinventing the wheel" (I hate that phrase.) your competitors will beat you to market.
I think I've covered two of the main points here.
1) It can be cheaper
2) It can be quicker to market.
Who cares if it requires a cluster of Opterons to run, software has always cost more than mass produced silicon waffers.
So, back to the main answer, which by now you probably can see, when to use EJB.
1) If you need distributed transactions.
2) If you need platform independance. (.Net clause)
3) If it will let you get to market quicker.
4) If you need declaritive security checks.
If none of those ifs apply to your project, then you would be an idiot riding hype if you use EJB. As for my EJB projects, I can generally say I need all 4. You can't force Linux on clients because they don't have administrative people that will patch it like they do windows. You can't use windows because it isn't reliable as a server generally. Let them pick and live with their own decision is my philosophy.
Still won't work... I would "bug" the lights near where emergency vehicles pass frequently and gather the info using IR as I drove by. Probably a 5$ disposable item, and I would collect the days key the first time it was used. You'ld have to put cameras and cops at lights to figure it out, then again, I might be able to do it from further away than just the intersection as well. I may even be able to make some buddies with the right people and set up the janitor with the system if he lets me make it. Also keep in mind that the bug I'm talking about would be about the size of dime as well. It would only be an ATMEL (one of the 8-pin chips), a LED, a photo transistor, and a coin cell. If it only transmits IR once a day, it would probably last at least 3 months, probably 6-12.
The best way to do it is the challenge method and RF, not IR. This would help track the progress of emergency vehicles as well and could be wired directly back to 911 so they can tell how far away the vehicle is. If you put one on your car, 911 tracking systems would see it, then you would get busted pretty fast. Nothing beats wiring the whole system up properly with security in mind to begin with. Eventually you would request a per incident key through hardwired land lines that are transmitted over using site keys that are carried out to the site so that loosing a key anywhere is a simple matter to fix.
It's not capillary action really... it is cohesion to the ball. The ball is rotated, and the ink would rather be on paper than the ball, so the ink leaves, thus creating a void which sucks in more ink.
There are some real space pens that don't use nitrogen pressure and can be used in both freezing cold and blistering heat because the ink is actually a near solid. The ball's friction tears off the ink as it goes, and the suction (the ball is mainly used as a valve in almost all ink pens) pulls ink in without letting air into the ink well.
There is a folk tale about ink pens and the cold war that I can't verify. It had to do with the US spending 5 million dollars or more in research to develop a pen that was cheap and would work in space while the Russians just used pencils. If anyone can find the origin of the folk tale, or actually find evidence for this story, I'ld love to read more.
I wonder what effects things like this has on the jewelry industry. In the next century, there will probably be children's toys that make and cut diamonds into any shape they can imagine with their 3d holographic editing system :)
Yeah, I'm interested in their magical switches too, because the network is always the bottleneck on a cluster. It's good we have a few people researching though.
:). I have some Pov Files I created that take me 8h to render on my little 3 machine cluster. You should try cheating to render a movie perhaps. When I was at EKU (this was before both the clusters at UK) we put together 15 486's as a test to see what kind of computer could be built with retro hardware from the trash can. We ended up having the a decent, modern "computer" for free.
I was just pointing out that if it's 40X the performance, and 130X the price, that's not two unheardof. Hopefully someone will do an Opteron cluster.
I just see similar comparisons in J2EE performance as well on clusters of machines. Someone will come out with the best price per performance offor for enterprise machines that doesn't scale beyone 2 machines and they claim that it is because the application server is superior.
BTW, I really love that you guys used Povray to test the cluster, even though any rendering that takes 2.5s is probably all network latency
I think clusters would be a lot easier if Linux supported equal cost multipath routing easily and effeciently. Using PVM to route requests is a little crazy. I wrote my own network communication most of the time.
Oh, another thing that probably helps the Big Mac is their support of a faster PCI bus (generally macs support both 64bit and 66mhz) which can limit some Gbit cards.
32bit*33mhz = 1Gbit to share between all the network cards (for single PCI bus systems). 4X the bandwidth I'm sure would help.
Yes, but you make the common mistake that we are working on a linear scale here. Back when they were testing a portion of Big Mac they were getting twice the effecience they are now. If you go look at the exact flops and divide the cluster in costs, you'll see that you are getting very close to the KASY0's performance/price ratio.
Technically, the highest performance/price ratio would probably be a cheap chinese/tiwanese watch, but those APEX DVD players probably aren't far behind. I guess if you really measured it all out, the best performance/price ratio is probably an XBox since MS is loosing money on them.
When the KASY0 is doing TFLOPS, then maybe you can compare the two.
(Not trying to knock UK, I have many friends that worked on that cluster... I live about 15 minutes away.)
Yeah, but it shouldn't still be classified, and that's the point. There are investigators working for Sci-Fi channel that make up scenario's every day, and they can't figure out why it would still be classified. At least I think we deserve some kind of reason why reports on a 1965 incident are still classified.
The only other plausible theory I've heard is that it might contain information that would damage foreign relations. Still they could release a report saying what the hell it was without giving away it's contents. They've done nothing thus far, so it's degraded into an issue only resolvable by a judge.
Then they have no reason to keep that classified, but it's still classified, and no one will let it go without a lawsuit apparantly!
They've actually researched enough and have some first hand accounts that it was not a U-2 crash. Even one person here on /. lived close to the incident and spoke to a firefighter that went out there before the others arrived, and described it as an acorn shaped object. They've elimenated the possiblity that it was a Satellite by talking to NASA themselves (RTA). You're theory is very unlikely. If you know anything about the Kecksburg incident, you'll know that civilians where held at gun point, and one reporter ended up dead. So, I think we have a right to know what the hell happened there at least to investigate the murder of a reporter.
/.'ers come post here so skeptical that they hate skeptics without even knowing their reasons. I hate most when an idiot laughes at UFO researchers saying they are loons without even seeing the evidence they have.
I just don't like the fact that
Think of it this way. If an alien landed today, and shook hands with the locals, would you believe the media? If not, you are severely biased the other way. Either admit that you don't know what happened, which means investigating is a good thing, or admit that you are a horribly biased bastard and don't post. Posting extreme skepticism is not helping any debate.
(This isn't directed completely at you blair1q, I just summed up my problem with about half of the +4 insightful posts here.)
This exact post is in every PHP/JAVA story on /., all the links are usually broken, and the writer is an idiot.
Java was written in C as well. Java compiles directly to machine byte-code on load (which happens once per class, which means only once per jsp file just like PHP).
My real question is how does this troll keep getting modded up?
Just for your information, I had to switch from PHP to Java because PHP was impossible to maintain.
Actually, that's his free speach. I could tell everyone where to buy drugs, and that would be perfectly legal. If you have a problem with copyright infringement, take it up with those infringing, besides, now you know where to find them.
Someone misused a word on Slashdot... Inconceivable!
I agree with you, developers generally don't talk to the customer. If you had actually talked to several customers (I have to deal with nearly 30 in my main project), I think you would get the point that the customer generally doesn't know what they want. After years of dealing with them, I've gotten a handle on heading off their requests with features before they can ask for them usually. It saves time in the long run if you design a flexible model to begin with; instead of scrapping portions of the project, adapt them. Make your model vague. My favorite is creating N:M relationships with the relation table containing an ID defining the relationship. That way you can relate different sets of objects in different ways, but the same code can show which objects are related and how.
Basically, I like to believe that XP is a few notches too far up on the design issue, but not because the customer knows what they want, but rather because you can design a flexible archetecture. I've only met one customer that knew what they wanted, and that was because they wanted a duplicate of something else, except they wanted it stable and fast.
yeah, but most of OpenGL is mapping textures stored on the card to pixels. If anything, this shows how a system built with this in mind (the root of the story) could indeed be much faster than it being a side-effect. The main point is that X has very, very little to do with OpenGL rendering that a game like Quake does.
The more likely scenario was that the remote machine was rendering the OpenGL locally, then sending it back to the client. The network would support up to 20fps at 640x480x16. If it had just loaded the textures onto the card on the local machine, then transferred only the Vectors, I'm sure we can both agree that 100fps would not be unreasonable for even a remote machine on a 100mbit network because there is no concievable way that it takes 100mbit* 1s/100fps=1mbit/f=976K per frame to transfer vertices and instructions. If you look at the benchmarks for cards, you'll see very quickly that bus speed to the card doesn't do much except let the game load textures onto the card quicker.
In fact, if someone wants to argue enough about this, I may be driven to enough rage to write a stub opengl.so to put in your quake3 directory that just sends the function calls through some home brew RPC protocol to a client X app on the other end. Hell, I bet I could use XML and still see full frame rates.
I never multitask during games by choice. Some applications demand attention. I've moved every one of those applications onto my cheap laptop. Let's me run Gaim (talk to people while I camp), Mozilla (for cheat codes of course), Mozilla Mail (in case something important actually does come up), and lots of other business-esque software so I can see who's calling (from the phone system) to see if I need to mute before I pick up.
:)
All the tweaking gamers out there should love this CD thing, for now. Then at some point they'll say, "Hey, you, Linux Geek! Can I copy this to my hard drive?" and of course you install a very compact Linux distro so that nothing else eats the precious CPU. Next you set up Mozilla for them so they don't have to reboot to browse the web. You set up Evolution, then they can get their mail. You set up StarOffice, and they can do their homework. Gamers won't want to reboot to windows except for the games that require it eventually (because they'll be slightly slower... at least that is true with some games like Unreal 1, and anything Quake). More demand for games on Linux (and therefore optimized drivers) is what it's all about, and I'm all for it. Now if NeverWinter would just conform to my theory that Linux is faster than Windows, I'ld be a very happy person!
Hey, I know some people, maybe I try to get Anand (of Anandtech) to run Linux benchmarks on cards to see who has the best Linux drivers. Someone did one back in 2000, but that's a lot dated. If it was in the benchmarks, I think it would help. (I'm very pleased with the NVidia drivers so far, but I haven't compared them to windows as much as I should. I think they use pretty much the same code because OpenGL is OpenGL even on Linux.)
Sorry, I've been writting(not righting) code for days straight, and haven't used English in a while. That part of my brain is choked for blood flow.
Do you really want geeks showing up on your doorstep in full chainmail, a sword, and the amulet of Yendor? I think you should very much encourage them to keep their fantasies stored as harmless electronic and magnetic charges far, far away from this "Real Life". I mean, have you ever actually been to a LARP (Live-Action Role Playing)? It's definately not pretty. I know a guy who ended up getting stitches in a hospital. I guess he thought he had more hit points than he actually did. That's ok, he eventually did end up back at town center, and all his digits were intact and functional after seeing the healer. I think he gave up adventuring though, and he now works answering other adventurers' questions about these magical devices that help in the creation of scrolls at Lexmark.
Now, do you really want us role playing geeks to invade real life? You know these people that role play are usually the ones that right your software. Next time you check your X11 log for the last error to see if it's complaining about the mouse or display modes and all you see is runes, you'll learn to appreciate the fact there is a time and place for these fantasies. I personally don't want to have to describe my problem on the mailing lists using Quenya.
I guess in your development model you write more than you read?
I bet you even are going to say that you use more transactions than you query.
Finding/Reading/Loading data is always going to be the key to the best performance. Transactions are second to that. 3000X faster at reading affords it a lot of breathing space when it comes to transactions because everyone visiting slashdot is reading, but only a few post. This is the real world model.
I'm skeptical when it comes to easily retrieving and indexing data for reporting. You need to be able to query aggregate (yes, this is easy enough to code) but you need to join related objects. Do I have to do my own query planning/joining? What's the transaction API like? Do I lock objects then update them? What about deadlock? (Easy enough to avoid by anyone smart enough to know any of the algorithms like having to sort the order in which you lock resources, but still a hassle.) So many questions!
I didn't want to do it yet, but I think I'm going to start running j2ssh. I don't know how stable it is at this point, does anyone know? I'll run it on different port and test it for a while I guess.
You can get the LGPL SSH2 server/client written in Java from http://3sp.com/products/sshtools/sshtools.php
Take a laser and move it side to side. Shine this on a mirror that is tilted not on the plane that the light is traveling and you'll see that it isn't side to side anymore.
Dishes weren't what I was talking about refracting the RF. Objects around the antenna like trees, cars, and houses seem to do a great job is dispersing signals. A lot of our signal before we used a dish was being bounced off of houses across the street then back to the target.
A dish may only be reflecting signal but it gets mostly signal directly from the point of origin because it reflects RF away that isn't striking the dish almost perpendicular to the plane that runs tangent to it's vertex.
It makes sense to me now that I realize thought that the 99% bulk of the signal a dish picks up is not bouncing off of houses and cars and the Earth or ionosphere. This is generally the case with every other antenna including Yagi style.