Your appeal to authority is based on references(Chris Hecker and John Carmack) who were commenting on the state of Direct3D over 10 years ago! Chris Hecker's article discusses execute buffers, a feature in Direct X version 3! John Carmack's.plan is from 1996. Comparing DirectX 3 and DX9 is just silly. The API has drastically changed. Right around DX9 it was doing a better job of exposing the underlying hardware than OGL. As of 2003(the last time I used OpenGL) OGL had all these vendor extensions, which meant you had to write different versions of your code(and your shaders!) to work on different hardware(e.g. NV_VERTEX_PROGRAM/NV_FRAGMENT_PROGRAM was nvidia only, and you needed different code for ATI). They finally came out with ARB extensions that worked on all hardware. Also both Nvidia and ATI had separate extensions for uploading geometry to the GPU for a while. It was a total pain.
Here's one random example where DirectX has more helper functions than OpenGL. DirectX has built in support for Quaternions: http://msdn.microsoft.com/en-us/library/bb281611(VS.85).aspx OpenGL does not(you would need to use a custom library). Built in font rendering is another example. You have to load bitmap fonts in OpenGL.
There is an instant search in the start menu, so you can type in the name of a program or file and it will appear in the start bar. No more hunting through menus(where did did starcraft go? Is it under games, blizzard, starcraft, etc?), and it makes it a lot easier to find your programs. This same search is applied to most of the windows configuration tools. For the control panel, you can also search(e.g., to turn off UAC), and it will filter out the non-relevant entries.
If you hover over programs in the task bar at the bottom of the screen, a preview of the window will render as a popup, which makes it easier to switch between multiple instances of the same program.
I've also found that the wireless networking management is much better than XP's.
>>This would be like a third party mirroring Slashdot's stories without Slashdot's advertisements, costing Slashdot revenue.
No it wouldn't. Slashdot makes money by selling advertising with the promise of eyeballs. RyanAir makes money by selling flights(and other services). This is more about whether they can force people to go through their portal(and thus sell their add-ons). They still make money selling flights. If the other crap was worth it, then maybe more people would buy it. If the flights are loss leaders for their other crap, then their business model is not going to survive the web.
Also, if it was illegal, as you assert, then why wouldn't they just get a legal injunction against these sites? It sounds like this has been going on for a while. The fact that they didn't pursue a legal solution, yet are trying to cause chaos for the sites suggests that they don't have a legal leg to stand on.
It works the other way too. Try doing any kind of dynamic class loading in c++ - welcome to COM! In contrast, Java and C# have built in mechanisms for dynamic class loading(and code generation).
I'd say go and look further at C# 3.0, with LINQ, lambdas, and expression trees. And it's all natively supported by the language and runtime, so you'll probably get a real compile error rather than template goo.
One more advantage of c# and java - since it doesn't have the archaic text #include, it can compile orders of magnitude faster than a c++ program. We're talking 2 minutes for 100k LOC. Heck, doing hello world in Boost and friends probably will take 2 minutes to compile on it's own to resolve all the template references:)
WTF are you talking about, your post is full of non-sequiters.
>>Expect a continued messy transition to 64 bit computing that will favor Intel, the other monopoly laggart. Vista and XP both shipped with 64 bit versions, specifically x86-64, which was developed originally by.....drumroll.....AMD! How exactly is ignoring IA64 for x86-64 favoring Intel?
>>Some nonsense about 32 bit computing Windows 95 was 32 bit software. Maybe you mean using a protected memory model and pre-emptive multitasking(which is an operating system concept and has nothing to do with application software). Even Mac OS 9 didn't have this, and WinNT(which predates it by several years) did!
>>Free software has handed M$ it's ass for 64 bit software and architecture independence. Guess what! Windows runs on more than x86: IA64, DEC Alpha and x86-64 come to mind as current and past platforms.
I think your point is that people don't have an incentive to buy a new computer or upgrade their operating system. You really need new killer apps to drive an upgrade cycle; lacking that, why should people upgrade?
Smart pointers have their own issues and still don't provably guarantee object lifetime. For instance I can hold a raw pointer to some object which is later destroyed by the smart pointer, and then use this destroyed resource.
c++ needs RAII for memory, which is the predominant resource. Java would only use it for e.g. file hooks, pipes, etc as garbage collection handles the memory issues. Furthermore, the disposable pattern can be used for these other resources. So destructors have value primarily to deallocate memory(and other resources), even in the face of exceptions...which is what try/finally is for.
Anyway, just because language x doesn't do things like c++, it doesn't mean that it's lacking. In non-c++ languages you don't tend to need to do "destroy _immediately_ while dropping last reference to."
>>(1) Java has no useful destructors because no object has predictable scope. try {... } finally {... }
Java does not do RAII the way c++ does, it does it differently. That doesn't make it broken, just slightly more verbose.
>>(2) Since everything is a pointer in Java, you have to bend over backwards to pass-by-value. The fact that the language doesn't even begin to provide copy-construction semantics. What a miserable PITA. Copy construction semantics typically don't make sense when applied to pointers(or references). Hence all of the nonsense about how to implement the copy constructor and assignment operator in c++. How many newbies have written c++ code that uses a default copy ctr, but news and deletes in the ctr and dtr, and then they wonder why their program crashes after making a copy of their object?
Java makes a better intro language than c++. At my university people were thrown into complex c++ projects to sink or swim...and most of them sank. That's because an intro class should lay the foundations for future work: it should teach you how to program. When you have a bunch of extra bullshit to deal with, that task becomes much more difficult. See C string manipulation for another example. Of course you should eventually learn c or c++ as part of a well rounded education. But java makes a great learning tool.
All spending is discretionary. Just because some law was written around WW2 doesn't mean that congress can't change how it works. This is a sad excuse to change the conversation. The taxes are different? They're still taxes paid by the same people! In fact, the middle class pays more towards social security because of the income cap on social security.
It is disingenuous to say that non-discretionary spending should be ignored.
>>With major corporations from the likes of Microsoft to IBM hiring principally outside the US in China and India, this is where the jobs will be and thus, where the grad students are coming from.
This is misleading. Microsoft has research centers in India and China, but the majority of the software engineering happens in Redmond. Windows, Office, SQL Server, Visual Studio, XBox, Live.com, MSN, are all in the US, and they're not going anywhere. Many of the new overseas jobs are in sales. Here's an article with some more info.
>>The real slap in the face of the whole thing is that said companies than have the audacity to complain that we don't have enough educated workers to provide a workforce here in America.
And how do you know that there aren't enough educated workers? Have you tried hiring thousands of software developers and testers? Microsoft is looking for the best talent in the world. Guess what - people born in the US are not magically smarter than the rest of the world. Statistically, you would expect that there would be many top software engineer applicants from non-US countries, especially since they tend to come to the US for their education.
That H1B issue may be true for some companies, but Microsoft is not one of them.
In theory, it is embarassingly parallel. However, there are some practical limitations that will prevent the performance of multi-core ray tracing from scaling linearly with the number of cores.
The biggest problem is memory bandwidth. Every pixel can shoot arbitrary rays into the scene, which can random access your scene geometry(and textures, and lighting, etc). Since all those cores are sharing the same memory subsystem, they're going to overload the memory bus and sit there stalling. Yes, each core may have its own cache, although that starts getting real expensive real quick if your cache is at all largish. Solving the memory problems will be important for multi-core to take off past 2/4/8 and onto 80+.
Of that list, the Group program manager, product unit manager, and engineering manager are people managers, while the rest are most likely individual contributors.
>>Nevertheless, that was a deliberate tradeoff for backwards compatibility
What does this mean exactly? You can have type safe generics AND backwards compatibility. My old non-generic code will still work in a language with generics. Maybe you wanted to pass ListofT to a method that takes List. That's really a different issue - it's NOT backwards compatibility, more like cross-compatibility. If you wanted backwards compatibility you probably should have written your method to take IList in the first place. Whoops. The so-called backwards compatibility is really an excuse for lack of proper runtime support of generics. But nice try to spin it as an advantage.
Not only is there the issue of 1 main core with 8 additional cores, the 8 additional cores don't have random access to main memory and are really only efficient at doing stream processing - they are tuned for vector operations. I'd agree that the GP(and GGGP) missed the point. The issue is not single core vs. multi-core - it's standard multi-core shared memory architecture vs. whack-ass-we-designed-it-ourselves-DSP-style-multi -core. Developers get the idea of multithreaded programming - it's hard, but they understand it. It gets much harder when you have to manually handle memory transfers for each of the threads to ensure they have the data they need to process.
Perhaps, to them, it is the best FPS ever. I hope you're being sarcastic when refering to Deus Ex as the best FPS ever. Not that Deus is a bad game or anything. It's such a subjective thing - everyone has their own preferences. What I've noticed is that people tend to claim the first polished game that they've played in one genre as the "best ever." Probably because their view of that genre is largely defined by that game.
>>He's just saying that these people pretend to be "hardcore" gamers because they think it makes them "cool".
Yes, and I disagreed with it. I think people bought Halo because of several reasons, mainly because it was a good game - but not to be "cool." Halo 2 was such a success because it was a highly anticipated sequel to a platform-defining game. It was accessible. And I don't buy the poser/hardcore gamer dichotomy. I think it's more about what people like. A lot of people like to play Halo, and if those people don't include the traditional hardcore gamer, that doesn't make it a lesser game.
There are plenty of old-school gamers that will buy an xbox, or a ps3. It's not about poser vs. hardcore, or old-school vs. newbie. It's about what system has the games you want to play. Look at your case - a self-described nintendo fanboy picked up a ps2 because it had the games you wanted to play.
>>Just look at all the new "gamers" who jumped on the Halo bandwagon who tout it as the best fps ever, when PC gamers have been fragging each other for years already with games that are better than Halo.
What's with the gamer snobbery? Are you a better gamer than they are because you played wolf3d? Is it possible that the however million people who bought halo actually like it as a game? That they play it because it is fun?
Just because something is popular doesn't make it bad.
>> The Wii seems to appeal more to the people who actually game because of the games With the implicit suggestion that the only "real" games are on the Wii, and all the games on xbox and playstation are "fake" games that people don't really enjoy, they just play them to look cool. The fact that most 3rd party games aren't on nintendo has nothing to do with it at all. Give me a break.
>>The above statement in the article contradicts his statement about Wii trailing significantly behind the others. If most of the crossover ownerships of consoles include the Wii, and the amount of crossover is decent, then you would think that the Wii would actually be significantly ahead of the others.
No, you wouldn't think that, because you don't know the relative proportions of single-console households. For instance, assume the numbers were 30% xbox, 30% ps3, 10% wii, 15% xbox + wii, 15% ps3 + wii, 0% xbox and ps3, and 0% all 3. These proportions are consistent with both statements: crossover is decent(1/3 of the market has > 2 systems), and all crossovers contain the wii. But wait - in our example, wii is still in last place with only 40% market penetration, compared to 45% each for xbox and ps3.
And there are multiple Jason Andersons that work at Microsoft(a simple search on $SEARCH_ENGINE will confirm this), so it would appear to be a fairly common name.
Finally, console success is very directly related to number of consoles sold, because video game companies need to make money to keep console alive(remember the dreamcast), and said companies make money in several ways: profit/loss on the sale of the console itself, profit on first party games, and royalties for 3rd party games and accessories. It should be obvious that the more consoles you have, the more games and accessories you can sell. Also, the more market share any console has, the greater the chance of attracting 3rd party developers.
Slashdot Mirror
Your appeal to authority is based on references(Chris Hecker and John Carmack) who were commenting on the state of Direct3D over 10 years ago! Chris Hecker's article discusses execute buffers, a feature in Direct X version 3! John Carmack's .plan is from 1996. Comparing DirectX 3 and DX9 is just silly. The API has drastically changed. Right around DX9 it was doing a better job of exposing the underlying hardware than OGL. As of 2003(the last time I used OpenGL) OGL had all these vendor extensions, which meant you had to write different versions of your code(and your shaders!) to work on different hardware(e.g. NV_VERTEX_PROGRAM/NV_FRAGMENT_PROGRAM was nvidia only, and you needed different code for ATI). They finally came out with ARB extensions that worked on all hardware. Also both Nvidia and ATI had separate extensions for uploading geometry to the GPU for a while. It was a total pain.
Here's one random example where DirectX has more helper functions than OpenGL. DirectX has built in support for Quaternions: http://msdn.microsoft.com/en-us/library/bb281611(VS.85).aspx
OpenGL does not(you would need to use a custom library). Built in font rendering is another example. You have to load bitmap fonts in OpenGL.
It will enter an infinite loop if days == 366 and IsLeapYear is true.
There is an instant search in the start menu, so you can type in the name of a program or file and it will appear in the start bar. No more hunting through menus(where did did starcraft go? Is it under games, blizzard, starcraft, etc?), and it makes it a lot easier to find your programs. This same search is applied to most of the windows configuration tools. For the control panel, you can also search(e.g., to turn off UAC), and it will filter out the non-relevant entries.
If you hover over programs in the task bar at the bottom of the screen, a preview of the window will render as a popup, which makes it easier to switch between multiple instances of the same program.
I've also found that the wireless networking management is much better than XP's.
>>This would be like a third party mirroring Slashdot's stories without Slashdot's advertisements, costing Slashdot revenue.
No it wouldn't. Slashdot makes money by selling advertising with the promise of eyeballs. RyanAir makes money by selling flights(and other services). This is more about whether they can force people to go through their portal(and thus sell their add-ons). They still make money selling flights. If the other crap was worth it, then maybe more people would buy it. If the flights are loss leaders for their other crap, then their business model is not going to survive the web.
Also, if it was illegal, as you assert, then why wouldn't they just get a legal injunction against these sites? It sounds like this has been going on for a while. The fact that they didn't pursue a legal solution, yet are trying to cause chaos for the sites suggests that they don't have a legal leg to stand on.
And here's the XKCD link.
It works the other way too. Try doing any kind of dynamic class loading in c++ - welcome to COM! In contrast, Java and C# have built in mechanisms for dynamic class loading(and code generation).
:)
I'd say go and look further at C# 3.0, with LINQ, lambdas, and expression trees. And it's all natively supported by the language and runtime, so you'll probably get a real compile error rather than template goo.
One more advantage of c# and java - since it doesn't have the archaic text #include, it can compile orders of magnitude faster than a c++ program. We're talking 2 minutes for 100k LOC. Heck, doing hello world in Boost and friends probably will take 2 minutes to compile on it's own to resolve all the template references
The subscription sales include console revenue... did you RYOFA?
WTF are you talking about, your post is full of non-sequiters.
>>Expect a continued messy transition to 64 bit computing that will favor Intel, the other monopoly laggart.
Vista and XP both shipped with 64 bit versions, specifically x86-64, which was developed originally by.....drumroll.....AMD! How exactly is ignoring IA64 for x86-64 favoring Intel?
>>Some nonsense about 32 bit computing
Windows 95 was 32 bit software. Maybe you mean using a protected memory model and pre-emptive multitasking(which is an operating system concept and has nothing to do with application software). Even Mac OS 9 didn't have this, and WinNT(which predates it by several years) did!
>>Free software has handed M$ it's ass for 64 bit software and architecture independence.
Guess what! Windows runs on more than x86: IA64, DEC Alpha and x86-64 come to mind as current and past platforms.
I think your point is that people don't have an incentive to buy a new computer or upgrade their operating system. You really need new killer apps to drive an upgrade cycle; lacking that, why should people upgrade?
I know this wasn't your point, but here's an interesting take on why you use the start button to shut down windows: Why do you have to click the Start button to shut down?
What is the sound of a joke being missed?
Smart pointers have their own issues and still don't provably guarantee object lifetime. For instance I can hold a raw pointer to some object which is later destroyed by the smart pointer, and then use this destroyed resource.
c++ needs RAII for memory, which is the predominant resource. Java would only use it for e.g. file hooks, pipes, etc as garbage collection handles the memory issues. Furthermore, the disposable pattern can be used for these other resources. So destructors have value primarily to deallocate memory(and other resources), even in the face of exceptions...which is what try/finally is for.
Anyway, just because language x doesn't do things like c++, it doesn't mean that it's lacking. In non-c++ languages you don't tend to need to do "destroy _immediately_ while dropping last reference to."
>>(1) Java has no useful destructors because no object has predictable scope. ... ...
try
{
}
finally
{
}
Java does not do RAII the way c++ does, it does it differently. That doesn't make it broken, just slightly more verbose.
>>(2) Since everything is a pointer in Java, you have to bend over backwards to pass-by-value. The fact that the language doesn't even begin to provide copy-construction semantics. What a miserable PITA.
Copy construction semantics typically don't make sense when applied to pointers(or references). Hence all of the nonsense about how to implement the copy constructor and assignment operator in c++. How many newbies have written c++ code that uses a default copy ctr, but news and deletes in the ctr and dtr, and then they wonder why their program crashes after making a copy of their object?
Java makes a better intro language than c++. At my university people were thrown into complex c++ projects to sink or swim...and most of them sank. That's because an intro class should lay the foundations for future work: it should teach you how to program. When you have a bunch of extra bullshit to deal with, that task becomes much more difficult. See C string manipulation for another example. Of course you should eventually learn c or c++ as part of a well rounded education. But java makes a great learning tool.
All spending is discretionary. Just because some law was written around WW2 doesn't mean that congress can't change how it works. This is a sad excuse to change the conversation. The taxes are different? They're still taxes paid by the same people! In fact, the middle class pays more towards social security because of the income cap on social security.
It is disingenuous to say that non-discretionary spending should be ignored.
Not really sure where you got your numbers from, as they don't actually match the pages you linked to.
>>With major corporations from the likes of Microsoft to IBM hiring principally outside the US in China and India, this is where the jobs will be and thus, where the grad students are coming from.
This is misleading. Microsoft has research centers in India and China, but the majority of the software engineering happens in Redmond. Windows, Office, SQL Server, Visual Studio, XBox, Live.com, MSN, are all in the US, and they're not going anywhere. Many of the new overseas jobs are in sales.
Here's an article with some more info.
http://seattletimes.nwsource.com/html/microsoft/2003836681_microsoftjobs15.html
>>The real slap in the face of the whole thing is that said companies than have the audacity to complain that we don't have enough educated workers to provide a workforce here in America.
And how do you know that there aren't enough educated workers? Have you tried hiring thousands of software developers and testers? Microsoft is looking for the best talent in the world. Guess what - people born in the US are not magically smarter than the rest of the world. Statistically, you would expect that there would be many top software engineer applicants from non-US countries, especially since they tend to come to the US for their education.
That H1B issue may be true for some companies, but Microsoft is not one of them.
In theory, it is embarassingly parallel. However, there are some practical limitations that will prevent the performance of multi-core ray tracing from scaling linearly with the number of cores.
The biggest problem is memory bandwidth. Every pixel can shoot arbitrary rays into the scene, which can random access your scene geometry(and textures, and lighting, etc). Since all those cores are sharing the same memory subsystem, they're going to overload the memory bus and sit there stalling. Yes, each core may have its own cache, although that starts getting real expensive real quick if your cache is at all largish. Solving the memory problems will be important for multi-core to take off past 2/4/8 and onto 80+.
IA64 was neither high performance nor straightforward.
At Microsoft, the Program Manager position is not a management position. Instead, it is the role of an engineer responsible for defining the features of a product. You can read more here, if you're interested: http://members.microsoft.com/careers/careerpath/te chnical/programmanagement.mspx
Of that list, the Group program manager, product unit manager, and engineering manager are people managers, while the rest are most likely individual contributors.
>>Nevertheless, that was a deliberate tradeoff for backwards compatibility
What does this mean exactly? You can have type safe generics AND backwards compatibility. My old non-generic code will still work in a language with generics. Maybe you wanted to pass ListofT to a method that takes List. That's really a different issue - it's NOT backwards compatibility, more like cross-compatibility. If you wanted backwards compatibility you probably should have written your method to take IList in the first place. Whoops. The so-called backwards compatibility is really an excuse for lack of proper runtime support of generics. But nice try to spin it as an advantage.
Not only is there the issue of 1 main core with 8 additional cores, the 8 additional cores don't have random access to main memory and are really only efficient at doing stream processing - they are tuned for vector operations. I'd agree that the GP(and GGGP) missed the point. The issue is not single core vs. multi-core - it's standard multi-core shared memory architecture vs. whack-ass-we-designed-it-ourselves-DSP-style-multi -core. Developers get the idea of multithreaded programming - it's hard, but they understand it. It gets much harder when you have to manually handle memory transfers for each of the threads to ensure they have the data they need to process.
Hewlett Packard is based in Silicon Valley. They may have an office in Portland, but the company's history is in California.
Perhaps, to them, it is the best FPS ever. I hope you're being sarcastic when refering to Deus Ex as the best FPS ever. Not that Deus is a bad game or anything. It's such a subjective thing - everyone has their own preferences. What I've noticed is that people tend to claim the first polished game that they've played in one genre as the "best ever." Probably because their view of that genre is largely defined by that game.
>>He's just saying that these people pretend to be "hardcore" gamers because they think it makes them "cool".
Yes, and I disagreed with it. I think people bought Halo because of several reasons, mainly because it was a good game - but not to be "cool." Halo 2 was such a success because it was a highly anticipated sequel to a platform-defining game. It was accessible. And I don't buy the poser/hardcore gamer dichotomy. I think it's more about what people like. A lot of people like to play Halo, and if those people don't include the traditional hardcore gamer, that doesn't make it a lesser game.
There are plenty of old-school gamers that will buy an xbox, or a ps3. It's not about poser vs. hardcore, or old-school vs. newbie. It's about what system has the games you want to play. Look at your case - a self-described nintendo fanboy picked up a ps2 because it had the games you wanted to play.
>>Just look at all the new "gamers" who jumped on the Halo bandwagon who tout it as the best fps ever, when PC gamers have been fragging each other for years already with games that are better than Halo.
What's with the gamer snobbery? Are you a better gamer than they are because you played wolf3d? Is it possible that the however million people who bought halo actually like it as a game? That they play it because it is fun?
Just because something is popular doesn't make it bad.
>> The Wii seems to appeal more to the people who actually game because of the games
With the implicit suggestion that the only "real" games are on the Wii, and all the games on xbox and playstation are "fake" games that people don't really enjoy, they just play them to look cool. The fact that most 3rd party games aren't on nintendo has nothing to do with it at all. Give me a break.
>>The above statement in the article contradicts his statement about Wii trailing significantly behind the others. If most of the crossover ownerships of consoles include the Wii, and the amount of crossover is decent, then you would think that the Wii would actually be significantly ahead of the others.
No, you wouldn't think that, because you don't know the relative proportions of single-console households. For instance, assume the numbers were 30% xbox, 30% ps3, 10% wii, 15% xbox + wii, 15% ps3 + wii, 0% xbox and ps3, and 0% all 3. These proportions are consistent with both statements: crossover is decent(1/3 of the market has > 2 systems), and all crossovers contain the wii. But wait - in our example, wii is still in last place with only 40% market penetration, compared to 45% each for xbox and ps3.
And there are multiple Jason Andersons that work at Microsoft(a simple search on $SEARCH_ENGINE will confirm this), so it would appear to be a fairly common name.
Finally, console success is very directly related to number of consoles sold, because video game companies need to make money to keep console alive(remember the dreamcast), and said companies make money in several ways: profit/loss on the sale of the console itself, profit on first party games, and royalties for 3rd party games and accessories. It should be obvious that the more consoles you have, the more games and accessories you can sell. Also, the more market share any console has, the greater the chance of attracting 3rd party developers.
Soma is not a random blogger, he is the VP of developer division. How did this get modded insightful?