This is why you use java. How many bugs have been found in Java's SSL libraries? None that I've ever heard of, and that's not because they're obscure. Both of these bugs could never exist in java. They would both throw exceptions when encountered, and those would be handled, rather than coredumping the program.
Lets face it guys, if you're not writing an OS, a game, or a calculation based app (lapack, etc...), you're insane if you're not writing in java. That's all there is to it. Why work twice as hard so you can have twice as many bugs and support half as many platforms? It just doesn't make sense.
OK, time for more insanity. All I want is a box that only runs java. Simple as that. Give it a microkernel and a JVM, push everything possible into the VM, and have it only run java, with all the policies set appropriately.
Then there are not buffer overflows, no root exploits, and when you double click something in an email it will run without disk access or net access, or basically anything else that could hurt you. Virii would immediately crash, anything else would work fine.
The real problem is not the executables, it's that they get so many damn permissions for no good reason. EVERY single executable should the first time it asks for disk access pop up one of those boxes saying what it is trying to do and do you want it to..
[allow once] [allow always] [allow always for this file] [disallow]....
Then the problem would be over. Your normal apps would get all they need (and nothing more) the first time they run, and spyware and malware would be almost impossible to slip through.
Much better to start everything out in a small sandbox and ask the user each time the boundaries need to be increased than to just let it do what it needs to and hope it isn't a virus.
Basically, if you could get everything you needed in java (web browser, email client, office suite [the most difficult task], jboss, mail server, databse [hsql], etc...) and you could netboot it then the people at work would have extremely safe and secure boxes where they could just do work, rather than always worrying about attack.
It'd also let you run without protected or virtual memory, and thus the 30% performance boost would just about wipe out the cost of doing everything in java. Most client machines shouldn't be swapping these days anyway.
1) everyone lives in the city. Country folk are already screwed regularly. If there was no need to even consider them, then they would be wholly marginalized. The current system of districts ensures that they at least have senators, even if the representatives are all owned by the cities. 80% of the US population is urban. Same thing goes for other groups. This prevents tyrrany of the majority.
2) Multiple parties cause a strange kindo of gridlock, in particular, spoiler parties can cause horrible damage. Look at what the green party did to the Dems in the last election, and what they're going to do again in November. Anyone who says that nader voters are half republicans didn't notice the huge drop in Kerry poll numbers when nader entered the race. Two parties does just fine, no issue loses out just because it is supported by multiple parties.
3) Though our current president is a joke, that is n't always the case. It's nice to have one person responsible for virtually any grievance you may have, someone who can't say "that's not my problem, that's because of...." The president has enough power that he's expected to be responsible for the general state of the country, but not enough (historically, before GW tried to be Caesar) to really screw us.
4) Winner take all elections are needed for president, just that simple. Congress is almost always so evenly divided that a winner take all election doesn't really matter.
5) There is no such thing as a fair election. This is mathematically proven to be the case. There is no optimal electoral system, the electoral college and winner take all elections we have are as good as any other. They come close to being optimal in the case of only two parties, and are far better with two parties than parlimentary style are with many parties.
Basically, what I'm trying to say is that you don't know what you're talking about.
1) CPU cycles (client). This is a very low cost for most clients. Should not be significant on the client side, ergo, cost approaches zero.
2) CPU cycles (server). This is generally fairly low, especially if you're doing any significant calculation as well. Servers are fairly fast now, should only rarely be a problem. If it is a problem, get a sun box with one of those crypto cards, problem solved. Cost here might be a little more than on the client, but shouldn't break the bank.
3) CPU cycles (network). The overhead of encryption on the network is negligible. This is simply not a problem.
4) Software complexity. Much of the software out there has it built in, you just have to use it. Java has SSL, free for anyone to use, web browsers have it, ftp clients generally have sftp, ditto for terminal clients, web servers all have it available. This is basically a non-issue unless you're rolling your own software, and if so, just use IPSec, or one of the above packages.
5) Key management, this is why we have CAs. In addition, not really needed. Wiretapping is easy, performing a man in the middle attack is much more difficult. It at least raises the bar radically, and enormously increases the odds of getting caught.
The basic idea is this, law enforcement should be hard. Just that simple. If it's hard, then people will reserve the sword of justice for serious crimes. If it's easy and cheap to prosecute people for Sodomy, sitting on milk crates, etc... then everyone has their own pet peeve that they'll want to enforce, and the budget will accomodate them all. If the FBI really wants your data, they'll break in and give you a keylogger keyboard. If they aren't willing to spend 10,000 to get your data, then it's probably just a fishing expidition, and not the legitimate domain of law enforcement anyway.
This is what crypto is for. It won't really protect your data if the government really wants it, but in order to get it they have to really want it. They can't just hoover it all up and see who might have broken a law today, for future reference should you ever run for office, etc....
OK, I'm officially the crackpot at work, because I tell them to USE ENCRYPTION! Encryption costs basically nothing, there's no reason why anyone would host a website that didn't use, or at least offer SSL. Everyone out there should use encryption for everything. Get rid of telnet, use ssh, and sftp rather than ftp. Use SSL on your websites, etc.... Encryption literally costs nothing, there is no advantage to not using it, WHY DON'T YOU USE IT? If you're too stupid to protect yourself when doing costs you nothing, then maybe you never really needed rights to begin with.
I just don't understand. It's so incredibly easy to protect your rights in this area, do you want someone else to do it for you. Clicking a button renders all their BS moot. With the effort you spend complaining you could solve the problem, it's just a button click away.
If you want untappable phones, use VPN to run your VOIP from another jurisdiction, simple as that.
you know, I have the same problem, paying for a phone line just to get dsl, that is, until I forgot to pay my bill and Verizon shut down the phone line. Funny thing is that the broadband still worked.
So.....Maybe you should just not pay your bill for the phone line, and eventually you'll only have DSL.
let us not forget who paid for many of those factories, shall we. The amount the US spent (over the last half a century or so) rebuilding things is only rivaled by the amount it spent destroying them.
I agree completely that C# kidnof sucks. They added every possible means for people to screw up, and removed all the impediments to failure that java thoughtfully put in, for instance...
1) No checked exceptions. This is not a feature, it is a bug.
2) Incomplete stack traces. You either have to throw the exception out of the method, or you don't get to know what called the method. So you can either recover or diagnose, never both.
3) Properties, quite possibly the worst idea ever. Yeah, lets make an arbitrarily complex and difficult function call look like a simple read, and make it impossible for anyone to ever know the difference.
4) Operator overloading, it was always a bad idea. I don't want to see a += operation that performs some heinous calculation. In real life + and - commute, do your + and - operators commute? If you don't know what this means, then you shouldn't be overloading operators, and you shouldn't be using code where others have overloaded operators.
6) Not platform independent, not even a little bit, why would you make a VM that isn't, you know, virtual.
7) A terrible native interface. I thought the hope was to leave COM behind. And if you want to call a native C function, well, spin the wheel and see which function gets called today. Java's tight binding is much better, and faster.
Basically, C# is just the latest vehicle for VB programmers to vomit forth sucky software. Java is much better, but I don't like a few of the things they're adding in the next version. Were the varargs really necessary? How many functions will use them, is it really THAT difficult to box data into an array (which is what happens anyway) for the three function calls into a varargs function that you will ever in your life make. Also, including a printf like printer seems like a bad idea, it was confusing an nasty in C (and was dropped in favor of streams in C++), it doesn't need to be brought back.
However, all this being said, C# is better than most of the nastiness that calls itself a language these days, but it's not hard to be better than VB.
Being from a rural area, and now living in NYC, allow me to chime in.
While there is no mass transit, and a taxi ride out to my house from pretty much anywhere would easily blow $100, and I don't live within 10 miles of anyone, there will be drunk driving.
Now, that being said, people should be horribly punished for being wildly over the limit (yeah, driving with a.25 should probably result in never driving again), but the punishment shouldn't be so severe for borderline cases. In addition, ban cell phones from the roads, the are a far worse problem.
Lets face it, some people are better drivers drunk than a soccer mom is stone sober. If we're going to punish people for reckless driving, lets do that, punishing people for this particular "sin" and not punishing all the fine upstanding soccer moms who never could drive safely doesn't seem like a good answer.
The answer is simple, wireless will never be a viable solution for lots of people needing lots of bandwidth, end of story.
Don't believe the long winded philosophers, useable spectrum is scarce. If everyone in the country wants to connect to the tower ten miles away at 100 megabit speeds, it's just not gonna happen. There is too little spectrum. Cell phone reception is bad enough, internet, forget about it. My wireless router is almost worthless due to interference from my neighbor's cordless phones or hair dryers, or whatever.
Basically, a single optical fiber can carry a huge volume of bandwidth, far beyond what radio will ever manage, especially considering that my fiber optic line doesn't produce or receive interference.
And before anyone claims that they could just open up more spectrum, let me head that one off at the pass. There is very little useful spectrum. Most spectrum has lots of interference on it, often from thing that humans make (our electrical gadgets) and sometimes from things we don't (cosmic microwave background). Most of what we have should be reserved for radars and important communications, not wasted on my grandmother because it's slightly easier for her to get 10 mb/sec over wireless than 1gb/sec over fiber.
Personally, I've always thought that open office should be written completely in java. It's supposed to be portable right? Well, then if it was written in java, there would be an OS X port of it today, not a year and a half from now, and it would run on anything.
In my opinion OOO cannot survive as a C++ program forever, it must transition to Java in order to get portability and faster development. And of course, anyone who says anything about performance will henceforth be known as an idiot. OOO is slow already, java is now pretty fast, and nobody cares about the performance of their office suite anyway. What they want is an office suite that works on every OS in existence, and has all the features they need, that's what java offers.
Microsoft will try a similar tactic with Office, by porting it into.NET. They would be crazy not to. Porting to.Net will wipe out about 3/4 of their security bugs (probably) and double the speed of development.
Look at Limewire. Their gui is pretty good. Also, netbeans can be a bit slow, but their gui isn't that bad either.
Of course I say this as a man who owns an OS X box, so I get the aqua L&F, in addition to plenty of hardware graphics acceleration.
Now, remind me again why the don't just program in java if they want cross platform features? I've seen the performance comparisons, and java and C# are roughly neck and neck. In my own code, java blows C# out of the water, but that's due to the way we use C#. Here's a hint, autoboxing is not a good feature, it just makes it too easy to write horribly slow code.
I would also like to point out that the sheer volume of code written in Java is staggering. A sizeable fraction (maybe 30%) of open source projects are in Java, and most of the code written in business today is java.That proportion would be higher if open source hackers weren't so in love with low level/obscure languages. Look at the jakarta project by apache. Most network services should be java (or.NET, if you insist) becaue they are so much harder to attack and exploit. In addition, the difficulty of maintaining multiple versions for multiple platforms (your network service does run on multiple platforms, right?) is greatly reduced.
In addition, if you feel the need to have a free VM, why not try to produce at GPLed JVM? You could probably get half the code from Sun, and the rest of the work would be much easier because you can always see (as in have others describe to you, so as to not corrupt the project) the sun source code for the JVM.
Last, and debatably least, I would like to point out that if C# is just barely keeping pace now, it will have a hard time in the future. Most of the advantages C# enjoys are due to its code format (tail recursion, etc...) and value types. Both of these advantages will be (have been) steadily erased by VM advances. Escape analysis will overnight turn value types from a performance advantage to a performance disadvantage, as the VM will be able to use value types wherever it is faster (and possible) to do so. Then programmer declared value types would become a liability.
Welcome to the world of photonic crystals. Though some problems are presented by the fact the the featuresize on your average chip is already down into the UV range, so photonic crystals would have to manage UV light, which is apparently much worse than normal light for these things. Not much is transparent to it and you tend to bleed out a lot.
However, that aside, photonic crystal waveguides can do a great job of bending light around radii roughly on the order of the wavelength of the light, and can even act as gates (electrical field from a 'gate' laser changes the epsion of the material and thus opens or closes the gate).
Now as soon as someone can find a way to make cheap UV sized photonic crystals that are transparent in that wavelength I think we've got it made.:-)
The only problem is that this won't scale so well, and can't go a whole lot smaller than our current level of electrical chips. Not a whole lot of materials are suitable once you get into the X-ray range, for physical reasons (electrons don't move fast enough).
Amen, I particularly like the utopic visions of hdtv, as if that was waiting for photonics all along. And jesus, a computer spanning the country, it's called the internet, welcome to 1985, now move along.
I can't stand it when half wit twits (even those with PHDs) decide that they need to claim that their new bose-einstein condensate can make a better blender. Grow the fuck up.
Seriously, all of these things will be useful someday, a rare few might even be useful soon (this one would probably be useful as soon as they can get it out the door for making high throughput optical networking) but do they really have to insult my intelligence by telling me that a particle collider will make fusion a reality? Seriously, this is just stupid.
Good job for the scientists who made the breakthrough, whoever wrote up that article needs to be drug out into the street and shot.
Electrons are well suited for this, when paired with holes. I think it was always imagined that in a trinary system the + state would be positive charge, the - state would be negative, and the 0 state would be neutral. Since a capacitor can hold a + or a - equally well, it's somewhat wasteful to not use this in memory at least, though it is not clear that modern semiconductors can be made to work in a charge symmetric environment like this.
That's nice, but for those of us who actually use our computers for something other than surfing for porn, XP is woefully insufficient. My XP box crashes about twice a week because I'm using it for actual work, and as soon as I cause a little memory churn it starts swapping out all the programs (even when it never even gets close to filling up its memory) and becomes dog slow.
My computer at home is almost identical (1.6 G5 w 1.25 GB RAM Vs. 2.4 P4 w/ 1 GB RAM) and it never has these problems, despite running far more programs, and often much more intensive ones. Face it, for real work, windows doesn't cut it.
In my experience, the chart making of open office is indeed fairly poor. If they could substantially improve that, then that alone would put them on much better footing.
The second area that needs work is much more specialized. We would love to use open office at work if it had RTD (Real Time Data) functionality. Basically a little command you can type into a cell that will cause a function call to called into a library of code to fetch data into the spreadsheet. A lot of people use this for a lot of things, and it would dramatically help if there was a good way to get data into the spreadsheets.
I have used the current interface to OpenOffice, and frankly it sucks. It's extremely slow, and feels very kludgy from java.
Basically, that's what it would take for us. RTD, better charting (handling lots of data without croaking), and a better java interface. If those three were delivered we'd drop excel in a second.
I seem to have misplaced my cray, as soon as it turns up I'll get right on that. In the meantime, no proposed theory is really supported by the evidence all that well yet, and so my armchair quarterbacking tells me that in light of the extreme complexity of current theories, odds are good that something simpler will win out in the end. The simplest explanation seems to be that we don't fully understand gravity, simple as that. That was after all the final explanation to the luminferous ether, perhaps we should begin with paths that worked in the past before diverging onto the exotic pet projects of the theorists. And yes, I took a class from Brian Greene, he's a smart guy, very mathy, still quite possibly wrong. His math is correct, but I'm not sure this is the real application of it. Just a hunch.
As a physics major, I actually agree. Dark matter reminds me not so subtly of the luminiferous ether of days past. What's that you say, immense mass, completely transparent and immaterial, clusters throughout the univers exactly the way a smudge in a telescope would (halos around objects, etc...), perhaps our instrumentation (and understanding) is a little off. It's easier to swallow that gravity isn't exactly 1/r^2 over huge distances than it is to believe that the universe is full of stuff that we can't see or feel (except at large distances) that clusters around normal matter in a manner suggestive of a severe rounding error.
And IAAP (I am a physicist, well, physics major at least). My $.02 (worth less every day bush is in office).
True, to an extent. However, most decent programmers (henceforth, this will be an appendix to my definition of decent) would like their software to run in the future, as well as the present. If you decide to use hack #8812 to get your software to run on XP, then (by my definition) you suck, as there is no reason to believe that it will run on any future (or past) versions of windows even.
A company that knows what it is doing, would also like its code to run in the future, so it can be used in future games, rather than needing to completely reinvent the wheel each time. A company that doesn't think this is a good idea is fairly foolish. I have worked for companies on both sides of this issue, the ones that take the afirmative stance are many times more profitable than those that take the negative one. Admitedly though, I am not a game developer. It is even more shameful in my line of work (custom applications) for someone to not follow this principle as usually performance isn't much of an issue (giving up half of the performance is fine, if it reduces the number of dependencies on outside hardware/software).
Basically what I'm saying is this.....
1) Use standards, especially standards that are widely available and not dependent on a single vendor (java/opengl vs directX/c#). Avoid non-standard extensions like the plague, though apparently that's hard to do in OpenGL right now....
2) Patch for deficiencies very carefully (check os version before executing the 'quirks' mode of a module, make sure the module has a 'non quirks' mode in case the software one day actually works correctly).
3) If code should be portable, verify this fact. Often a recompile on a different architecture will flush out lots of bugs.
4) This is generally not applicable to games, but should be at least considered. Any code that can be written in java should be. Java is a lingua franca of computing code because it is basically the most strict language out there in common use. Porting from java to basically anything is easy, compiling java to basically anything is also easy. Anything written in java is guaranteed to still be useable ten years from now with only minimal alteration, if any is needed at all. Native code will almost never be useable in ten years, it'll be a constant struggle to keep it working after every release of os/library software. Don't undertake this battle unless you need to.
Once again, all of these rules are less applicable to games, where sometimes you have to do crazy things to get the proper performance, but a lot of software houses (most of them in fact) do this sort of thing even when they don't have to, that is my point.
I couldn't agree more. Any game that isn't available on more than one platform immediately at release is just a sign of shitty programmers.
Flat out, if you can't program in a way that would be portable (and not take about two years to port to a new platform) then get up, pack up your MSCE, admit that you're an idiot, and go into marketing where you belong.
Also, if you're a boss who thinks that adding 5% to development costs in order to make a cross platform game that ACTUALLY HAS A PRAYER OF WORKING ON A COMPUTER 5 YEARS FROM NOW is a waste of time and money, the please either get out of the industry, or tell me where you work so I can never buy anything from you.
Most mac games are vastly inferior to their windows counterparts, at least as far as performance goes. Sim City 4 runs better on my girlfriend's year and a half old laptop than it does on my G5, and that's just completely unacceptable (not that it's that bad on the G5 though). There is no way on earth that her laptop has even half the horsepower of my G5, in any subsystem, so clearly maxis just slapped together a half assed mac version and kicked it out the door.
Basically, my advice to those out there looking for mac games is to look no further than Blizzard. Get anything else, and it'll probably be way too slow. Actually, any game that is released simultaniously on mac and PC will be ok, but anything that goes through a long porting process, don't waste your time, it's going to suck.
Here's another example, Tropico. The performance of tropico (under OS 9 or OS X) is terrible beyond words. The game hangs at the slightest provocation, and horribly crashes OS X ALL THE TIME. In fact, with 10.3.2 it crashes the OS so badly every time you start it up that you can't get back to the desktop, it somehow corrupts the video system, so though the rest of the computer (the BSD part) continues to run correctly, you can't see anything on the monitor even after quitting the game.
All I have to say is let the buyer beware. And in addition, we really need reviewers who can bring themselves to comment on HORRIBLE stability bugs. I always read the reviews before getting a game, and they never mention bugs, even if the game is so buggy that it doesn't hardly run, the reviewer will never mention that little tidbit. Don't you think that is maybe more relevant than anything else?
Slashdot Mirror
This is why you use java. How many bugs have been found in Java's SSL libraries? None that I've ever heard of, and that's not because they're obscure. Both of these bugs could never exist in java. They would both throw exceptions when encountered, and those would be handled, rather than coredumping the program.
Lets face it guys, if you're not writing an OS, a game, or a calculation based app (lapack, etc...), you're insane if you're not writing in java. That's all there is to it. Why work twice as hard so you can have twice as many bugs and support half as many platforms? It just doesn't make sense.
my $.02
OK, time for more insanity. All I want is a box that only runs java. Simple as that. Give it a microkernel and a JVM, push everything possible into the VM, and have it only run java, with all the policies set appropriately.
....
Then there are not buffer overflows, no root exploits, and when you double click something in an email it will run without disk access or net access, or basically anything else that could hurt you. Virii would immediately crash, anything else would work fine.
The real problem is not the executables, it's that they get so many damn permissions for no good reason. EVERY single executable should the first time it asks for disk access pop up one of those boxes saying what it is trying to do and do you want it to..
[allow once] [allow always] [allow always for this file] [disallow]
Then the problem would be over. Your normal apps would get all they need (and nothing more) the first time they run, and spyware and malware would be almost impossible to slip through.
Much better to start everything out in a small sandbox and ask the user each time the boundaries need to be increased than to just let it do what it needs to and hope it isn't a virus.
Basically, if you could get everything you needed in java (web browser, email client, office suite [the most difficult task], jboss, mail server, databse [hsql], etc...) and you could netboot it then the people at work would have extremely safe and secure boxes where they could just do work, rather than always worrying about attack.
It'd also let you run without protected or virtual memory, and thus the 30% performance boost would just about wipe out the cost of doing everything in java. Most client machines shouldn't be swapping these days anyway.
That system would suck, here's why.....
...." The president has enough power that he's expected to be responsible for the general state of the country, but not enough (historically, before GW tried to be Caesar) to really screw us.
1) everyone lives in the city. Country folk are already screwed regularly. If there was no need to even consider them, then they would be wholly marginalized. The current system of districts ensures that they at least have senators, even if the representatives are all owned by the cities. 80% of the US population is urban. Same thing goes for other groups. This prevents tyrrany of the majority.
2) Multiple parties cause a strange kindo of gridlock, in particular, spoiler parties can cause horrible damage. Look at what the green party did to the Dems in the last election, and what they're going to do again in November. Anyone who says that nader voters are half republicans didn't notice the huge drop in Kerry poll numbers when nader entered the race. Two parties does just fine, no issue loses out just because it is supported by multiple parties.
3) Though our current president is a joke, that is n't always the case. It's nice to have one person responsible for virtually any grievance you may have, someone who can't say "that's not my problem, that's because of
4) Winner take all elections are needed for president, just that simple. Congress is almost always so evenly divided that a winner take all election doesn't really matter.
5) There is no such thing as a fair election. This is mathematically proven to be the case. There is no optimal electoral system, the electoral college and winner take all elections we have are as good as any other. They come close to being optimal in the case of only two parties, and are far better with two parties than parlimentary style are with many parties.
Basically, what I'm trying to say is that you don't know what you're talking about.
Ok, I'm replying to my own.... here goes.....
1) CPU cycles (client). This is a very low cost for most clients. Should not be significant on the client side, ergo, cost approaches zero.
2) CPU cycles (server). This is generally fairly low, especially if you're doing any significant calculation as well. Servers are fairly fast now, should only rarely be a problem. If it is a problem, get a sun box with one of those crypto cards, problem solved. Cost here might be a little more than on the client, but shouldn't break the bank.
3) CPU cycles (network). The overhead of encryption on the network is negligible. This is simply not a problem.
4) Software complexity. Much of the software out there has it built in, you just have to use it. Java has SSL, free for anyone to use, web browsers have it, ftp clients generally have sftp, ditto for terminal clients, web servers all have it available. This is basically a non-issue unless you're rolling your own software, and if so, just use IPSec, or one of the above packages.
5) Key management, this is why we have CAs. In addition, not really needed. Wiretapping is easy, performing a man in the middle attack is much more difficult. It at least raises the bar radically, and enormously increases the odds of getting caught.
The basic idea is this, law enforcement should be hard. Just that simple. If it's hard, then people will reserve the sword of justice for serious crimes. If it's easy and cheap to prosecute people for Sodomy, sitting on milk crates, etc... then everyone has their own pet peeve that they'll want to enforce, and the budget will accomodate them all. If the FBI really wants your data, they'll break in and give you a keylogger keyboard. If they aren't willing to spend 10,000 to get your data, then it's probably just a fishing expidition, and not the legitimate domain of law enforcement anyway.
This is what crypto is for. It won't really protect your data if the government really wants it, but in order to get it they have to really want it. They can't just hoover it all up and see who might have broken a law today, for future reference should you ever run for office, etc....
OK, I'm officially the crackpot at work, because I tell them to USE ENCRYPTION! Encryption costs basically nothing, there's no reason why anyone would host a website that didn't use, or at least offer SSL. Everyone out there should use encryption for everything. Get rid of telnet, use ssh, and sftp rather than ftp. Use SSL on your websites, etc.... Encryption literally costs nothing, there is no advantage to not using it, WHY DON'T YOU USE IT? If you're too stupid to protect yourself when doing costs you nothing, then maybe you never really needed rights to begin with.
I just don't understand. It's so incredibly easy to protect your rights in this area, do you want someone else to do it for you. Clicking a button renders all their BS moot. With the effort you spend complaining you could solve the problem, it's just a button click away.
If you want untappable phones, use VPN to run your VOIP from another jurisdiction, simple as that.
you know, I have the same problem, paying for a phone line just to get dsl, that is, until I forgot to pay my bill and Verizon shut down the phone line. Funny thing is that the broadband still worked.
So.....Maybe you should just not pay your bill for the phone line, and eventually you'll only have DSL.
let us not forget who paid for many of those factories, shall we. The amount the US spent (over the last half a century or so) rebuilding things is only rivaled by the amount it spent destroying them.
I agree completely that C# kidnof sucks. They added every possible means for people to screw up, and removed all the impediments to failure that java thoughtfully put in, for instance...
1) No checked exceptions. This is not a feature, it is a bug.
2) Incomplete stack traces. You either have to throw the exception out of the method, or you don't get to know what called the method. So you can either recover or diagnose, never both.
3) Properties, quite possibly the worst idea ever. Yeah, lets make an arbitrarily complex and difficult function call look like a simple read, and make it impossible for anyone to ever know the difference.
4) Operator overloading, it was always a bad idea. I don't want to see a += operation that performs some heinous calculation. In real life + and - commute, do your + and - operators commute? If you don't know what this means, then you shouldn't be overloading operators, and you shouldn't be using code where others have overloaded operators.
6) Not platform independent, not even a little bit, why would you make a VM that isn't, you know, virtual.
7) A terrible native interface. I thought the hope was to leave COM behind. And if you want to call a native C function, well, spin the wheel and see which function gets called today. Java's tight binding is much better, and faster.
Basically, C# is just the latest vehicle for VB programmers to vomit forth sucky software. Java is much better, but I don't like a few of the things they're adding in the next version. Were the varargs really necessary? How many functions will use them, is it really THAT difficult to box data into an array (which is what happens anyway) for the three function calls into a varargs function that you will ever in your life make. Also, including a printf like printer seems like a bad idea, it was confusing an nasty in C (and was dropped in favor of streams in C++), it doesn't need to be brought back.
However, all this being said, C# is better than most of the nastiness that calls itself a language these days, but it's not hard to be better than VB.
Being from a rural area, and now living in NYC, allow me to chime in.
While there is no mass transit, and a taxi ride out to my house from pretty much anywhere would easily blow $100, and I don't live within 10 miles of anyone, there will be drunk driving.
Now, that being said, people should be horribly punished for being wildly over the limit (yeah, driving with a
Lets face it, some people are better drivers drunk than a soccer mom is stone sober. If we're going to punish people for reckless driving, lets do that, punishing people for this particular "sin" and not punishing all the fine upstanding soccer moms who never could drive safely doesn't seem like a good answer.
The answer is simple, wireless will never be a viable solution for lots of people needing lots of bandwidth, end of story.
Don't believe the long winded philosophers, useable spectrum is scarce. If everyone in the country wants to connect to the tower ten miles away at 100 megabit speeds, it's just not gonna happen. There is too little spectrum. Cell phone reception is bad enough, internet, forget about it. My wireless router is almost worthless due to interference from my neighbor's cordless phones or hair dryers, or whatever.
Basically, a single optical fiber can carry a huge volume of bandwidth, far beyond what radio will ever manage, especially considering that my fiber optic line doesn't produce or receive interference.
And before anyone claims that they could just open up more spectrum, let me head that one off at the pass. There is very little useful spectrum. Most spectrum has lots of interference on it, often from thing that humans make (our electrical gadgets) and sometimes from things we don't (cosmic microwave background). Most of what we have should be reserved for radars and important communications, not wasted on my grandmother because it's slightly easier for her to get 10 mb/sec over wireless than 1gb/sec over fiber.
Personally, I've always thought that open office should be written completely in java. It's supposed to be portable right? Well, then if it was written in java, there would be an OS X port of it today, not a year and a half from now, and it would run on anything.
In my opinion OOO cannot survive as a C++ program forever, it must transition to Java in order to get portability and faster development. And of course, anyone who says anything about performance will henceforth be known as an idiot. OOO is slow already, java is now pretty fast, and nobody cares about the performance of their office suite anyway. What they want is an office suite that works on every OS in existence, and has all the features they need, that's what java offers.
Microsoft will try a similar tactic with Office, by porting it into
Look at Limewire. Their gui is pretty good. Also, netbeans can be a bit slow, but their gui isn't that bad either. Of course I say this as a man who owns an OS X box, so I get the aqua L&F, in addition to plenty of hardware graphics acceleration.
Now, remind me again why the don't just program in java if they want cross platform features? I've seen the performance comparisons, and java and C# are roughly neck and neck. In my own code, java blows C# out of the water, but that's due to the way we use C#. Here's a hint, autoboxing is not a good feature, it just makes it too easy to write horribly slow code.
I would also like to point out that the sheer volume of code written in Java is staggering. A sizeable fraction (maybe 30%) of open source projects are in Java, and most of the code written in business today is java.That proportion would be higher if open source hackers weren't so in love with low level/obscure languages. Look at the jakarta project by apache. Most network services should be java (or
In addition, if you feel the need to have a free VM, why not try to produce at GPLed JVM? You could probably get half the code from Sun, and the rest of the work would be much easier because you can always see (as in have others describe to you, so as to not corrupt the project) the sun source code for the JVM.
Last, and debatably least, I would like to point out that if C# is just barely keeping pace now, it will have a hard time in the future. Most of the advantages C# enjoys are due to its code format (tail recursion, etc...) and value types. Both of these advantages will be (have been) steadily erased by VM advances. Escape analysis will overnight turn value types from a performance advantage to a performance disadvantage, as the VM will be able to use value types wherever it is faster (and possible) to do so. Then programmer declared value types would become a liability.
Just my $.02, now worth $.016 thanks to Bush.
You missed HP-UX, and I imagine you also missed a whole slew of *BSDs (I think Net and Open also have ports, like FreeBSD).
Welcome to the world of photonic crystals. Though some problems are presented by the fact the the featuresize on your average chip is already down into the UV range, so photonic crystals would have to manage UV light, which is apparently much worse than normal light for these things. Not much is transparent to it and you tend to bleed out a lot. However, that aside, photonic crystal waveguides can do a great job of bending light around radii roughly on the order of the wavelength of the light, and can even act as gates (electrical field from a 'gate' laser changes the epsion of the material and thus opens or closes the gate). Now as soon as someone can find a way to make cheap UV sized photonic crystals that are transparent in that wavelength I think we've got it made. :-)
The only problem is that this won't scale so well, and can't go a whole lot smaller than our current level of electrical chips. Not a whole lot of materials are suitable once you get into the X-ray range, for physical reasons (electrons don't move fast enough).
Amen, I particularly like the utopic visions of hdtv, as if that was waiting for photonics all along. And jesus, a computer spanning the country, it's called the internet, welcome to 1985, now move along.
I can't stand it when half wit twits (even those with PHDs) decide that they need to claim that their new bose-einstein condensate can make a better blender. Grow the fuck up.
Seriously, all of these things will be useful someday, a rare few might even be useful soon (this one would probably be useful as soon as they can get it out the door for making high throughput optical networking) but do they really have to insult my intelligence by telling me that a particle collider will make fusion a reality? Seriously, this is just stupid.
Good job for the scientists who made the breakthrough, whoever wrote up that article needs to be drug out into the street and shot.
Electrons are well suited for this, when paired with holes. I think it was always imagined that in a trinary system the + state would be positive charge, the - state would be negative, and the 0 state would be neutral. Since a capacitor can hold a + or a - equally well, it's somewhat wasteful to not use this in memory at least, though it is not clear that modern semiconductors can be made to work in a charge symmetric environment like this.
That's nice, but for those of us who actually use our computers for something other than surfing for porn, XP is woefully insufficient. My XP box crashes about twice a week because I'm using it for actual work, and as soon as I cause a little memory churn it starts swapping out all the programs (even when it never even gets close to filling up its memory) and becomes dog slow. My computer at home is almost identical (1.6 G5 w 1.25 GB RAM Vs. 2.4 P4 w/ 1 GB RAM) and it never has these problems, despite running far more programs, and often much more intensive ones. Face it, for real work, windows doesn't cut it.
In my experience, the chart making of open office is indeed fairly poor. If they could substantially improve that, then that alone would put them on much better footing. The second area that needs work is much more specialized. We would love to use open office at work if it had RTD (Real Time Data) functionality. Basically a little command you can type into a cell that will cause a function call to called into a library of code to fetch data into the spreadsheet. A lot of people use this for a lot of things, and it would dramatically help if there was a good way to get data into the spreadsheets. I have used the current interface to OpenOffice, and frankly it sucks. It's extremely slow, and feels very kludgy from java. Basically, that's what it would take for us. RTD, better charting (handling lots of data without croaking), and a better java interface. If those three were delivered we'd drop excel in a second.
I seem to have misplaced my cray, as soon as it turns up I'll get right on that. In the meantime, no proposed theory is really supported by the evidence all that well yet, and so my armchair quarterbacking tells me that in light of the extreme complexity of current theories, odds are good that something simpler will win out in the end. The simplest explanation seems to be that we don't fully understand gravity, simple as that. That was after all the final explanation to the luminferous ether, perhaps we should begin with paths that worked in the past before diverging onto the exotic pet projects of the theorists. And yes, I took a class from Brian Greene, he's a smart guy, very mathy, still quite possibly wrong. His math is correct, but I'm not sure this is the real application of it. Just a hunch.
As a physics major, I actually agree. Dark matter reminds me not so subtly of the luminiferous ether of days past. What's that you say, immense mass, completely transparent and immaterial, clusters throughout the univers exactly the way a smudge in a telescope would (halos around objects, etc...), perhaps our instrumentation (and understanding) is a little off. It's easier to swallow that gravity isn't exactly 1/r^2 over huge distances than it is to believe that the universe is full of stuff that we can't see or feel (except at large distances) that clusters around normal matter in a manner suggestive of a severe rounding error. And IAAP (I am a physicist, well, physics major at least). My $.02 (worth less every day bush is in office).
True, to an extent. However, most decent programmers (henceforth, this will be an appendix to my definition of decent) would like their software to run in the future, as well as the present. If you decide to use hack #8812 to get your software to run on XP, then (by my definition) you suck, as there is no reason to believe that it will run on any future (or past) versions of windows even.
A company that knows what it is doing, would also like its code to run in the future, so it can be used in future games, rather than needing to completely reinvent the wheel each time. A company that doesn't think this is a good idea is fairly foolish. I have worked for companies on both sides of this issue, the ones that take the afirmative stance are many times more profitable than those that take the negative one. Admitedly though, I am not a game developer. It is even more shameful in my line of work (custom applications) for someone to not follow this principle as usually performance isn't much of an issue (giving up half of the performance is fine, if it reduces the number of dependencies on outside hardware/software).
Basically what I'm saying is this.....
1) Use standards, especially standards that are widely available and not dependent on a single vendor (java/opengl vs directX/c#). Avoid non-standard extensions like the plague, though apparently that's hard to do in OpenGL right now....
2) Patch for deficiencies very carefully (check os version before executing the 'quirks' mode of a module, make sure the module has a 'non quirks' mode in case the software one day actually works correctly).
3) If code should be portable, verify this fact. Often a recompile on a different architecture will flush out lots of bugs.
4) This is generally not applicable to games, but should be at least considered. Any code that can be written in java should be. Java is a lingua franca of computing code because it is basically the most strict language out there in common use. Porting from java to basically anything is easy, compiling java to basically anything is also easy. Anything written in java is guaranteed to still be useable ten years from now with only minimal alteration, if any is needed at all. Native code will almost never be useable in ten years, it'll be a constant struggle to keep it working after every release of os/library software. Don't undertake this battle unless you need to.
Once again, all of these rules are less applicable to games, where sometimes you have to do crazy things to get the proper performance, but a lot of software houses (most of them in fact) do this sort of thing even when they don't have to, that is my point.
I couldn't agree more. Any game that isn't available on more than one platform immediately at release is just a sign of shitty programmers.
Flat out, if you can't program in a way that would be portable (and not take about two years to port to a new platform) then get up, pack up your MSCE, admit that you're an idiot, and go into marketing where you belong.
Also, if you're a boss who thinks that adding 5% to development costs in order to make a cross platform game that ACTUALLY HAS A PRAYER OF WORKING ON A COMPUTER 5 YEARS FROM NOW is a waste of time and money, the please either get out of the industry, or tell me where you work so I can never buy anything from you.
Here's my take on this matter....
Most mac games are vastly inferior to their windows counterparts, at least as far as performance goes. Sim City 4 runs better on my girlfriend's year and a half old laptop than it does on my G5, and that's just completely unacceptable (not that it's that bad on the G5 though). There is no way on earth that her laptop has even half the horsepower of my G5, in any subsystem, so clearly maxis just slapped together a half assed mac version and kicked it out the door.
Basically, my advice to those out there looking for mac games is to look no further than Blizzard. Get anything else, and it'll probably be way too slow. Actually, any game that is released simultaniously on mac and PC will be ok, but anything that goes through a long porting process, don't waste your time, it's going to suck.
Here's another example, Tropico. The performance of tropico (under OS 9 or OS X) is terrible beyond words. The game hangs at the slightest provocation, and horribly crashes OS X ALL THE TIME. In fact, with 10.3.2 it crashes the OS so badly every time you start it up that you can't get back to the desktop, it somehow corrupts the video system, so though the rest of the computer (the BSD part) continues to run correctly, you can't see anything on the monitor even after quitting the game.
All I have to say is let the buyer beware. And in addition, we really need reviewers who can bring themselves to comment on HORRIBLE stability bugs. I always read the reviews before getting a game, and they never mention bugs, even if the game is so buggy that it doesn't hardly run, the reviewer will never mention that little tidbit. Don't you think that is maybe more relevant than anything else?
Out of curiosity, did you report/fix the bugs?