Why vow something stupid like that? Companies have turned around before (ahem AMD!), and remember, Cyrix is under a new owner. When it comes out, read the reviews. If it is faster and cheaper, buy it! No reason to make a vow against it!
Actually, back in the heyday of the 6x86 (or maybe 5x86), the fastest pentium out was the vinilla Pentium sans MMX. Back then the performance of Cyrix chips (both FP and Int) were unmatched. Then the Pentium MMX came and brought out a really massive FPU, and Cyrix was left in the dust. The core architecture of the 6x86 is actually faster in integer apps than an equivilant Pentium MMX (maybe even PII) because a 233 MHz Cyrix performs like a 233MHz MMX for integer, but only runs at like 208 MHz or something.
Actually, it does not suck. For most things that consumers do (ie. games, word processing, most media stuff) having a full speed on die L2 of 256K is better than having half speed 512K L2, or 1 or 2 meg bus speed L3. True, for servers its slower, but even for 3D studio MAX, rendering times increase less than 10 percent going from a 512K PIII to a 1meg Zeon. For consumer apps, going to 256K (like the new coppermine PIIIs) on die full speed is acutally about 10-20% faster than the old half speed on card 512K L2.
Re:Will Apple finally see OpenSource light at last
on
MacOS X DP3
·
· Score: 1
Yes, but you neglect the point that if D3D hadn't gotten as competitive with GLIDE, we would still be using 3Dfx cards! Yes it was a mistake to not push OpenGL 3D for PC, but you must remember, this was before open source open gl, and writing an ICD took a long time.
Re:Will Apple finally see OpenSource light at last
on
MacOS X DP3
·
· Score: 1
Are you deaf? I know GL was designed for acceleration. My point was that it never would have been accelerated on the PC unless MS had introduced Direct3D, pushing developers to use hardware acceleration.
Are you kidding that windows is unusable for C development? A. Games are much harder on the system than C development. Games are large things that use 99% of the proc 100% of the time you're running it. C development will use 100% for the short bursts that you're compiling. B. What kind of computer are you running? (Or are you trying SC3K, Halflife, and the sims at the same time!) I play a lot of Unreal and even the unstable beta patches have not yet crashed on me. I also am a C++ programmer. I write a lot of DirectDraw programs, and aside from programming errors (not my fault, the DirectX API is to frail about bad parameters) the thing has yet to crash on me. True, Windows is not super stable, but it is hardly unstable enough to say that its unusable for C (or C++) development. The guy above has a lot of good points. X IS too bloated. Here you have this beautiful (for a UNIX) lean kernel, then on top of it you run fat, ugly piglike things such as X and GNOME and Enlightenment. Just sad, to tell the truth. PS> Intra-thread messaging systems should not be buffered. They share the same address space and thus getting the message across as fast as possible should be the primary goal. Unlike inter-process messaging, the performance hit of buffering outweighs the overhead of sending more message packets.
Re:How much of Direct X was written by Microsoft?
on
MacOS X DP3
·
· Score: 1
Actually, DirectDraw evolved from an Intel technology. I do know that the core DirectX services were made in-house, DirectDraw, DirectInput, DirectSound. Some of the newer ones might have been acquired, like DirectMusic or DirectAnimation, but even then, they would have had to go through a major overhaul to have the same API as the rest of DirectX. There is nothing else like the DirectX API (mostly for good reasons, it is a bitch to program, but a dream once you get it running.)
Okay, here is my requisite bitch session about the problems in X. First. Is it just me, or is it really stupid to put a transparent networking protocol at the heart of the system. Something like that should really be at a higher level. I like the SunView approach. The windowing system would directly use ioctl()s to access the graphics card. Who's up for a port of THAT to Linux? Second. What has been done to lighten up X. When I hear statements like, "Netscape uses up even more memory than X!" its not to reassuring, considering that most non-unix windowing systems are fairly lightweight. Believe it or not, Quartz is based on the concept of a "lightweight window process" and if you've ever used BeOS, you know how lightweight it is. Its integrated into the app server (I think) and its memory usage barly registers. And GNOME doesn't help much either. Do any of the people on the dev list have any indication of X4's memory usage in comparison to X3.3.x? May they pull a Be and actually DECREASE memory usage in a major feature release? Third. How good is DRI? Is it up to par with the cool SGI push X out of the way tricks? Is Linux OpenGL support finally going to whoop NT's? I hope so.
What is wrong with you Linux people? Since when is "decent" ever good enough when it comes to speed?
Re:Will Apple finally see OpenSource light at last
on
MacOS X DP3
·
· Score: 1
I would seriously like to denounce you for saying that Microsoft has led the the perpetual reinvention of the wheel. Sure on/. its the phat thing to say, but it is a lie. MS HAS innovated. They have put out a lot of crappy products, but the HAVE innovated. Take Direct3D for example. If they hadn't brought Direct3D to the market, everyone would be using Glide and 3Dfx would be a monopoly and the great nVidia cards wouldn't exist. That was innovation. That was good for the industry. However, later, when OpenGL started getting accelerated, instead of killing of D3D, they played all these tricks to get D3D have bigger support. Yes, that was bad for the industry, but much less than if D3D had never been written. They also invented the rest of DirectX. DirectSound 3D is the reason that today you can buy any 3D sound card on the market and not worry about it being compatible with your games. If Direct Sound had never been invented, we would have Aureal as the monopoly in sound. Same thing with force feed back and iForce. When I was shopping for a computer in 1996, I knew what I had to get. A Voodoo and the Monster Sound A3D. There wasn't any choice, because everyone used Glide and A3D. I didn't like the Monster Sound because it didn't have good MIDI, but I had no other choices. When MS came in and innovated, the consumer got the huge range of products available today. Sure they have made an assload of bad desicions and bad products, but they still don't overshadow everything thay have done for the computer industry. You may not like them, you may not like Bill Gates, you may not like Windows, but I sure bet that you're happy that you're sitting there with you Sound Blaster Live! running you're GeForce 256 with your force feedback joystick humming away. They all wouldn't exist without MS.
Nobody should have to enfore it. Nobody enforces they fact that X is the default windowing system. If an API were constructed that would seperate WM from the rest of the system, then I think people would flock to it. People like programming Be and Win32 because its so integrated. It cuts down on bloat and makes the most of existing code. I doubt the KDE guys want software to be incompatible with GNOME and I doubt developers want to have to deal with dozens of layers of libraries. I think they wish they could spend more time perfecting the UI, but they can't because the foundatin isn't there. If they don't want to, fine. But why are people programming for GNOME and KDE? It offers them an integrated development environment and provides a wide array of services. Having a system API does not require someone to be in charge. It does not require the taking away of freedom. Windows has a system API, but there are dozens of other ones. If I don't like MFC (which sucks by the way) I can use OWL. Or I could ignore Win32 altogether and use SDL or the cygwin environment. In the end, all I'm saying is that having all these seperate APIs degrades Linux's ease of use, performance and stability. Building a common foundation for higher level stuff doesn't take away freedom. Just like using OSS as the sound system, or POSIX as the low level system API doesn't take away from freedom. Its a practical necessity.
Look, is it my fault if the UNIX guys can't get their act together? Actually, this API could be a standard POSIX library and thus portable to ANY Unix system.
My beef, as you put it, is with the fact that if a KDE app wants to run under GNOME, you have to have all the KDE libraries installed, and they have to load into memory. Thus you have TWO bloated libraries in memory. I could care less about the widgets. Those SHOULD be different and replacable. But Qt is not just a set of pretty widgets. It defines and entire API for interacting with the system and provides many services to the app. THAT should be in a standard API. Widgets and WMs are fine, but is there a need for a print arhitecture in a widget set?
What is up with all the people calling Microsoft facist? They are a business. Their sole purpose in life is to make money. Thats what businesses do, thats what they've been doing since they were invented. Any fair way for them to make money is fine. Thats the whole concept of capitalism. Now, you can argue capitalism vs. the world all you want, but thats pointless. True, some of Microsoft's actions are unfair, and yes their software is pretty bad, but consider this, what are their intentions? Their tactic of "embracing" and extending is a beautiful business plan. It has made them a ton of money. If you don't like the fact that it put wennie companies out of business, then thats your problem. If you want and alternative, then write it yourself! And thats what the open source community has done. They don't bitch about MS, they shut up and code. But there are still those jackasses who think there is something ethically wrong with Microsoft. Granted, their practice or not giving discouts to those OEMs who did not abide by their wishes is unfair, but thats the only real fault I can find with them. Many of you also forget that the companies that MS put out of business are better gone than here. Its called evolution. The good companies, EA, Inuit, Adobe, etc, are still going strong, mainly because their products kick ass! MS hasn't been able to beat them. However, Netscape et al, have been beaten. Mainly because their products weren't good enough. Have you used communicator lately? A web browser should not take longer than Office to start up! Also, enough bitching about Windows and its problems. It can't be made more stable than it is now. Win9x is based on DOS, and they can't get rid of DOS support even if they want to. People would be in an uproar. NT is new, and if you've ever used it, its a damn good OS. On my computer, 300MHz 192 meg RAM, NT 4 is faster than win98 and just as fast as Linux. (I use Photoshop, 3D studio and VC++) It still has the best OpenGL support of any OS available on x86. But I digress. You have the ardant NT users pointing to the Linux users calling them commies. You have the ardant Linux users pointing to MS and calling it facist. But in truth the only facist commies are the guys who think their way is the "One True Way"(TM)
I hear a lot of people on one side say that Linux should have a standard window manager, and a lot of people on the other side saying that choice is good. There is a perfectly good compromise between the two, but it cannot be achieved because the GNOME and KDE developers are, to put it nicely, brain-dead. Both KDE and GNOME have very nice widget sets, Qt and GTK, respectivly. The problem is that they provide an API, not just widgets, so you have to utterly unrelated libraries tied together. The reason people are pushing for a standard interface in mainly app compatibility. Choice is fine, but when one desktop does not nativly run the apps of another desktop, then its not the same OS anymore. They are adopting the emacs train of thought, thinking that they ARE the OS. Sorry, but there is no reason that a widget set should have printing services. The best thing to do would be this. Make a Linux API. It has Posix, but something to access the higher level system would be nice. Yes, X has this to an extent, but it is a pain in the ass to program to all these different layers. Adopt either Qt or GTK, (preferably GTK) and strip out all the widget stuff, and flesh out the rest of it so it can function as a full API for the system. Then make a strict API for the system API to talk to the window manager, and move all the widget stuff into there. Thus what you end up with is a system that is almost as customizable as the current system, but without the hassles of having to deal with so many APIs. Policy should be seperate of implementation. Thus an app could ask for a window, do operations to the window, etc, all through the system API, but the look and feel would be governed by whatever window manager one had installed.
Not really, if you consider they can have multiple busses going to memory. (IE. You read the post correctly) They have a 2 1K busses and a 512 bit texture bus. Some simple elementary (or middle or high, depending on what state you're in) school addition and you get, wow! 2560!!! Isn't math FUN!
I don't know what you've been reading, but I've known about Willamette since before I bought my PII300 back in '98. Its been on the design table since even before then.
Actually, all 2 or 3 gigabytes of the Win32 SDK, sample programs, libraries, headers and all is available for the cost of a 3 day download from MS's website. Its a hell of a lot better than anything for linux I'll tell you that. They have 400 pages of documentation for DirectDraw alone!
I don't know why all so many people are complaining that MS is just doing this to benifet themselves. OF COURSE thats why they'd do it. They're a business. There sole purpose for existance is to make MONEY. They aren't people who do this on their free time and have day jobs. If they just Open sourced windows to benefit the computing community, people would like them more, but it would kill their business. Pure Open Source is not making anyone money. Sure Linux has some obscene portion of the server market, but is Redhat making billions of dollars like you'd expect? No, I think there still in the low hundred millions, or lower. And why not? Open Source is okay for the software, but it is a dumb-ass business model. A large company can buy one copy of Linux and install it on all their servers. Half the time they won't even need support from Redhat because their own IT guys will be able to fix a lot of the problems. On the desktop, you have make even less money. The average windows user spends something like 5 bucks a year on support, most of it on long distance calls to Seattle. I personally have never called technical support, and most windows users don't. The less knowledageable asks a friend because they wouldn't under stand tech-support anyway. Business users call their IT guy, and knowledagble people look it up on the internet. So your market for making money is reduced to the 1% who know enough to follow orders from the techsupport guy, is willing to pay for support (these days, most support in the form of e-mail is free) instead of asking a friend, and doesn't know enough to find it on the internet. 10 years ago you'd have been laughed out of business school if you'd come up with the "charge for support, not the product" model!
Alright, the list time/. posted a kernel release on the board, I defended it saying that an occasional heads up was important for some projects. But this is going a bit far. I was fine with 2.3.41 because it had been a while since the last post, but here's another only.05 releases later? Sure a lot of dev releases are important, but here are a few suggestions that could fix the kernel dev release problem. 1. Definatly announce full releases. 2. Announce and pre- builds like 2.2.0pre9 3. Post about major feature additions. 4. Announce the occasional dev build just to keep people aware of the project. It would also be more helpful if, instead of saying, here is another dev kernel, people would post about some of the stuff in the kernel, bug-fixes, new feautures, articles about new subsystems, etc. Not everyone is on the kerneldev mailing list you know.
Whoosh, totaly over your head wasn't it? Its funny how/.ers think normal people code video games. Clue: Video games are coded by corperations, by teams of people. They are infernally attached to custom code. Even though C was available for the Playstation since its introduction, people wrote entire games in ASM for the longest time. Its true that the 32K does not include hardware access libraries, but think about it. What libraries are those? You basically have a set of functions that wrap directly around the hardware. No memory protection so no need to go through the kernel. Developer makes a request, you hit write value x to memory location y and call interupt z, and boom, you're done. The lack of need for a driver model of any sort frees you from half the job right there. Don't expect the PS2 libraries to get much more complex. All it will probably be is a TCP/IP stack. Any services you want to implement will be up to you. And why should it be anything but that? If I can to support fancy stuff I should code it myself and not burden everyone else who just wants the fastest stream of data possible. The only place I can see it getting more complex are some services for Firewire and some TCP/IP services. Not even a network API, just some functions to get a TCP/IP connection running. And I doubt developers want WinCE (wince, get it?) functionality at the cost of speed. Gaming is very competitive and programming more or less directly to the hardware give you a big edge. It might not matter for someone making a browser for the system, but it does to the guy making the game. Developers don't do it to show of their coding skills. They do it because A) Its easy enough. Only have to write one set of custom code, and no maintainability issues. Once a set of code for a general type of game has been written, then the company re-uses it for future projects. But no need to make a general purpose OS to handle everything, an OS suitable for a driving game would suck when you ran the latest RPG! B) Its fast as a hell. Crash3 warp would bog down a 133MHz computer with a Voodoo card. Consider this is running on a 33MHz MIPS proc. with a 1994 era graphics chip. It has transparancy, something playstation isn't technically supposed to be able to do. I know you Linux people like abstraction and code reuse and "services at cost of performance" but thats not how its done in gaming. People expect games to continue to get better on the same exact hardware and expect 100% bug free programs. Alex St. John once said that computer users pay a 80% OS tax on everything and its true. If it really doesn't help, why bother putting an GPOS on the system?
This is almost too funny to be true. Stack up these two machines. PS2: -300 MHz processor -32 meg RAM -no HD X-Box -1000Mhz proc -64 meg RAM -4 gig HD X-Box wins right? Wrong! When you consider that the 300MHz proc in the PS2 is made to run games and only games, you realize that it wipes the floor of even the 1GHz proc. Then you realize, hey! the Windows derived OS will take up half the RAM! Playstation's OS is about 32K and probabaly won't get over 512K in PS2. Okay fine the HD is there. That way MS can issue patches for the system. Then trow in the fact that the PS2 graphics processer whoops the ass of even the GeForce (theoretical 75M/triangles per second vs. theoretical 15M/triangles per second) you see who will trump who. I thing that MS really was really dreamig when they thought this one up. A. If they can it it in under $300 I will print out the slash code and eat it. B. If they can get people who are used to one crash in 5 years and bug-free games to live with a crash a week and endless patches, then have the government sue them for using mind control. MS is in way over its head on this one.
I think there are a few points that will go over most Slashdotter's heads. The first one that I would like to clear up is... DREAMCAST DOES NOT RUN WINCE. A version of it has been made, the web browser uses it I think, but thats about the extent of it. A developer would be stupid to run WinCE, or ANY general purpose OS for that matter, on gaming hardware. Another thing I think a lot of people miss is that this is probably will not run the Windows derived OS for games, just different apps like browsers and such. On a console, the OS is just a bunch of support libraries wrapped around a real time kernel. In the Playstation for example, the OS is just 32K, not just the kernel, the whole thing. Remember, in a console, the system doesn't really need most of the services that other systems need. A network API? Just make a direct interface to TCP/IP. Graphics drivers? Why? Just write some ASM libraries to access graphics hardware directly. Memory protection? Who needs it? Virtual memory, maybe necessary cuz of the HD, but greatly simplified due to the fact that only one process is running. Which brings up another key point. The OS on a console does no governing whatsoever. There is not contention between process for resources. If it is needed, the app accesses it directly. As I remember, most of the first playstation games were written in ASM for the large part. And if you think running Linux on this bad boy is a good idea, all I ask is why? UNIX is a general purpose OS and a very abstracted one at that. Sure it may be cool to get it running, but will it be useful? Ultimatly I think it is a bad idea for general purpose OSs to try to get on game consoles. You're talking about a market where stability is not really key, but people are used to it. I'm not talking UNIX stability either. I have had my N64 for about 5 years. In that time, only one game has frozen on me. One! What is that like a 1600 day uptime? I think you can say that because this machine was being used at 100% for 2 or 3 (not just me, my brother) hours a day on average for 1600 days, easily the load of the linux email server sitting in the corner for 3 years.
You're so wrong I don't even know when to start. Yo, some one moderate this guy down for spreading FUD. I would, but I feel like posting on this thread. First. Win2K is nowhere near 11 years old. Cairo started just before windows 95 came out, around, '94. I know cuz I have the BYTE announcement for it. Cairo is not NT. Cairo was the code name for NT 5.0 which becamse W2K. Second, 95 is not a fork of the 3.1 code base. It was re-written for the most part with most existing code moved into protected mode modules. Its less like the evolution from kernel 2.0 to 2.2 and more like BSD to Next. Next put in a new core and put all the BSD cruft on top of that. Second, NT shares no kernel source with VMS, they don't have the right to. The only VMS stuff NT has is that the designers were originally working on VMS and brought a lot of design (but not code) into NT. I think you have Cario mixed up with NT. Cario is just a version of the OS formarly known as Windows NT, which is itself probably not dated '89. NT was from scratch, but desgin concepts were carried over from VMS. It is a microkernel design which is pretty clean at the core. If you haven't read about the NT kernel, you should, it is pretty nice. Its just the servers on top of the kernel are pieces of shit.
He forgets the counterpoint. If there is only Satan, then the christian goes to hell, and the devil worshippers go to heaven. Similarly if there is only the christian or muslim god (the two religions that believe one who is not one of them will go to hell, then everyone on the wrong side will go to hell. When your talking about something as abstract as theology, you have to be careful because normally logical cases don't apply.
Slashdot Mirror
Why vow something stupid like that? Companies have turned around before (ahem AMD!), and remember, Cyrix is under a new owner. When it comes out, read the reviews. If it is faster and cheaper, buy it! No reason to make a vow against it!
Actually, back in the heyday of the 6x86 (or maybe 5x86), the fastest pentium out was the vinilla Pentium sans MMX. Back then the performance of Cyrix chips (both FP and Int) were unmatched. Then the Pentium MMX came and brought out a really massive FPU, and Cyrix was left in the dust. The core architecture of the 6x86 is actually faster in integer apps than an equivilant Pentium MMX (maybe even PII) because a 233 MHz Cyrix performs like a 233MHz MMX for integer, but only runs at like 208 MHz or something.
Actually, it does not suck. For most things that consumers do (ie. games, word processing, most media stuff) having a full speed on die L2 of 256K is better than having half speed 512K L2, or 1 or 2 meg bus speed L3. True, for servers its slower, but even for 3D studio MAX, rendering times increase less than 10 percent going from a 512K PIII to a 1meg Zeon. For consumer apps, going to 256K (like the new coppermine PIIIs) on die full speed is acutally about 10-20% faster than the old half speed on card 512K L2.
Yes, but you neglect the point that if D3D hadn't gotten as competitive with GLIDE, we would still be using 3Dfx cards! Yes it was a mistake to not push OpenGL 3D for PC, but you must remember, this was before open source open gl, and writing an ICD took a long time.
Are you deaf? I know GL was designed for acceleration. My point was that it never would have been accelerated on the PC unless MS had introduced Direct3D, pushing developers to use hardware acceleration.
Are you kidding that windows is unusable for C development? A. Games are much harder on the system than C development. Games are large things that use 99% of the proc 100% of the time you're running it. C development will use 100% for the short bursts that you're compiling. B. What kind of computer are you running? (Or are you trying SC3K, Halflife, and the sims at the same time!) I play a lot of Unreal and even the unstable beta patches have not yet crashed on me. I also am a C++ programmer. I write a lot of DirectDraw programs, and aside from programming errors (not my fault, the DirectX API is to frail about bad parameters) the thing has yet to crash on me. True, Windows is not super stable, but it is hardly unstable enough to say that its unusable for C (or C++) development. The guy above has a lot of good points. X IS too bloated. Here you have this beautiful (for a UNIX) lean kernel, then on top of it you run fat, ugly piglike things such as X and GNOME and Enlightenment. Just sad, to tell the truth.
PS> Intra-thread messaging systems should not be buffered. They share the same address space and thus getting the message across as fast as possible should be the primary goal. Unlike inter-process messaging, the performance hit of buffering outweighs the overhead of sending more message packets.
Actually, DirectDraw evolved from an Intel technology. I do know that the core DirectX services were made in-house, DirectDraw, DirectInput, DirectSound. Some of the newer ones might have been acquired, like DirectMusic or DirectAnimation, but even then, they would have had to go through a major overhaul to have the same API as the rest of DirectX. There is nothing else like the DirectX API (mostly for good reasons, it is a bitch to program, but a dream once you get it running.)
Okay, here is my requisite bitch session about the problems in X. First. Is it just me, or is it really stupid to put a transparent networking protocol at the heart of the system. Something like that should really be at a higher level. I like the SunView approach. The windowing system would directly use ioctl()s to access the graphics card. Who's up for a port of THAT to Linux? Second. What has been done to lighten up X. When I hear statements like, "Netscape uses up even more memory than X!" its not to reassuring, considering that most non-unix windowing systems are fairly lightweight. Believe it or not, Quartz is based on the concept of a "lightweight window process" and if you've ever used BeOS, you know how lightweight it is. Its integrated into the app server (I think) and its memory usage barly registers. And GNOME doesn't help much either. Do any of the people on the dev list have any indication of X4's memory usage in comparison to X3.3.x? May they pull a Be and actually DECREASE memory usage in a major feature release?
Third. How good is DRI? Is it up to par with the cool SGI push X out of the way tricks? Is Linux OpenGL support finally going to whoop NT's? I hope so.
What is wrong with you Linux people? Since when is "decent" ever good enough when it comes to speed?
I would seriously like to denounce you for saying that Microsoft has led the the perpetual reinvention of the wheel. Sure on /. its the phat thing to say, but it is a lie. MS HAS innovated. They have put out a lot of crappy products, but the HAVE innovated. Take Direct3D for example. If they hadn't brought Direct3D to the market, everyone would be using Glide and 3Dfx would be a monopoly and the great nVidia cards wouldn't exist. That was innovation. That was good for the industry. However, later, when OpenGL started getting accelerated, instead of killing of D3D, they played all these tricks to get D3D have bigger support. Yes, that was bad for the industry, but much less than if D3D had never been written. They also invented the rest of DirectX. DirectSound 3D is the reason that today you can buy any 3D sound card on the market and not worry about it being compatible with your games. If Direct Sound had never been invented, we would have Aureal as the monopoly in sound. Same thing with force feed back and iForce. When I was shopping for a computer in 1996, I knew what I had to get. A Voodoo and the Monster Sound A3D. There wasn't any choice, because everyone used Glide and A3D. I didn't like the Monster Sound because it didn't have good MIDI, but I had no other choices. When MS came in and innovated, the consumer got the huge range of products available today. Sure they have made an assload of bad desicions and bad products, but they still don't overshadow everything thay have done for the computer industry. You may not like them, you may not like Bill Gates, you may not like Windows, but I sure bet that you're happy that you're sitting there with you Sound Blaster Live! running you're GeForce 256 with your force feedback joystick humming away. They all wouldn't exist without MS.
Nobody should have to enfore it. Nobody enforces they fact that X is the default windowing system. If an API were constructed that would seperate WM from the rest of the system, then I think people would flock to it. People like programming Be and Win32 because its so integrated. It cuts down on bloat and makes the most of existing code. I doubt the KDE guys want software to be incompatible with GNOME and I doubt developers want to have to deal with dozens of layers of libraries. I think they wish they could spend more time perfecting the UI, but they can't because the foundatin isn't there. If they don't want to, fine. But why are people programming for GNOME and KDE? It offers them an integrated development environment and provides a wide array of services. Having a system API does not require someone to be in charge. It does not require the taking away of freedom. Windows has a system API, but there are dozens of other ones. If I don't like MFC (which sucks by the way) I can use OWL. Or I could ignore Win32 altogether and use SDL or the cygwin environment. In the end, all I'm saying is that having all these seperate APIs degrades Linux's ease of use, performance and stability. Building a common foundation for higher level stuff doesn't take away freedom. Just like using OSS as the sound system, or POSIX as the low level system API doesn't take away from freedom. Its a practical necessity.
Look, is it my fault if the UNIX guys can't get their act together? Actually, this API could be a standard POSIX library and thus portable to ANY Unix system.
My beef, as you put it, is with the fact that if a KDE app wants to run under GNOME, you have to have all the KDE libraries installed, and they have to load into memory. Thus you have TWO bloated libraries in memory. I could care less about the widgets. Those SHOULD be different and replacable. But Qt is not just a set of pretty widgets. It defines and entire API for interacting with the system and provides many services to the app. THAT should be in a standard API. Widgets and WMs are fine, but is there a need for a print arhitecture in a widget set?
What is up with all the people calling Microsoft facist? They are a business. Their sole purpose in life is to make money. Thats what businesses do, thats what they've been doing since they were invented. Any fair way for them to make money is fine. Thats the whole concept of capitalism. Now, you can argue capitalism vs. the world all you want, but thats pointless. True, some of Microsoft's actions are unfair, and yes their software is pretty bad, but consider this, what are their intentions? Their tactic of "embracing" and extending is a beautiful business plan. It has made them a ton of money. If you don't like the fact that it put wennie companies out of business, then thats your problem. If you want and alternative, then write it yourself! And thats what the open source community has done. They don't bitch about MS, they shut up and code. But there are still those jackasses who think there is something ethically wrong with Microsoft. Granted, their practice or not giving discouts to those OEMs who did not abide by their wishes is unfair, but thats the only real fault I can find with them. Many of you also forget that the companies that MS put out of business are better gone than here. Its called evolution. The good companies, EA, Inuit, Adobe, etc, are still going strong, mainly because their products kick ass! MS hasn't been able to beat them. However, Netscape et al, have been beaten. Mainly because their products weren't good enough. Have you used communicator lately? A web browser should not take longer than Office to start up! Also, enough bitching about Windows and its problems. It can't be made more stable than it is now. Win9x is based on DOS, and they can't get rid of DOS support even if they want to. People would be in an uproar. NT is new, and if you've ever used it, its a damn good OS. On my computer, 300MHz 192 meg RAM, NT 4 is faster than win98 and just as fast as Linux. (I use Photoshop, 3D studio and VC++) It still has the best OpenGL support of any OS available on x86. But I digress. You have the ardant NT users pointing to the Linux users calling them commies. You have the ardant Linux users pointing to MS and calling it facist. But in truth the only facist commies are the guys who think their way is the "One True Way"(TM)
I hear a lot of people on one side say that Linux should have a standard window manager, and a lot of people on the other side saying that choice is good. There is a perfectly good compromise between the two, but it cannot be achieved because the GNOME and KDE developers are, to put it nicely, brain-dead. Both KDE and GNOME have very nice widget sets, Qt and GTK, respectivly. The problem is that they provide an API, not just widgets, so you have to utterly unrelated libraries tied together. The reason people are pushing for a standard interface in mainly app compatibility. Choice is fine, but when one desktop does not nativly run the apps of another desktop, then its not the same OS anymore. They are adopting the emacs train of thought, thinking that they ARE the OS. Sorry, but there is no reason that a widget set should have printing services. The best thing to do would be this. Make a Linux API. It has Posix, but something to access the higher level system would be nice. Yes, X has this to an extent, but it is a pain in the ass to program to all these different layers. Adopt either Qt or GTK, (preferably GTK) and strip out all the widget stuff, and flesh out the rest of it so it can function as a full API for the system. Then make a strict API for the system API to talk to the window manager, and move all the widget stuff into there. Thus what you end up with is a system that is almost as customizable as the current system, but without the hassles of having to deal with so many APIs. Policy should be seperate of implementation. Thus an app could ask for a window, do operations to the window, etc, all through the system API, but the look and feel would be governed by whatever window manager one had installed.
Not really, if you consider they can have multiple busses going to memory. (IE. You read the post correctly) They have a 2 1K busses and a 512 bit texture bus. Some simple elementary (or middle or high, depending on what state you're in) school addition and you get, wow! 2560!!! Isn't math FUN!
I don't know what you've been reading, but I've known about Willamette since before I bought my PII300 back in '98. Its been on the design table since even before then.
Actually, all 2 or 3 gigabytes of the Win32 SDK, sample programs, libraries, headers and all is available for the cost of a 3 day download from MS's website. Its a hell of a lot better than anything for linux I'll tell you that. They have 400 pages of documentation for DirectDraw alone!
I don't know why all so many people are complaining that MS is just doing this to benifet themselves. OF COURSE thats why they'd do it. They're a business. There sole purpose for existance is to make MONEY. They aren't people who do this on their free time and have day jobs. If they just Open sourced windows to benefit the computing community, people would like them more, but it would kill their business. Pure Open Source is not making anyone money. Sure Linux has some obscene portion of the server market, but is Redhat making billions of dollars like you'd expect? No, I think there still in the low hundred millions, or lower. And why not? Open Source is okay for the software, but it is a dumb-ass business model. A large company can buy one copy of Linux and install it on all their servers. Half the time they won't even need support from Redhat because their own IT guys will be able to fix a lot of the problems. On the desktop, you have make even less money. The average windows user spends something like 5 bucks a year on support, most of it on long distance calls to Seattle. I personally have never called technical support, and most windows users don't. The less knowledageable asks a friend because they wouldn't under stand tech-support anyway. Business users call their IT guy, and knowledagble people look it up on the internet. So your market for making money is reduced to the 1% who know enough to follow orders from the techsupport guy, is willing to pay for support (these days, most support in the form of e-mail is free) instead of asking a friend, and doesn't know enough to find it on the internet. 10 years ago you'd have been laughed out of business school if you'd come up with the "charge for support, not the product" model!
Alright, the list time /. posted a kernel release on the board, I defended it saying that an occasional heads up was important for some projects. But this is going a bit far. I was fine with 2.3.41 because it had been a while since the last post, but here's another only .05 releases later? Sure a lot of dev releases are important, but here are a few suggestions that could fix the kernel dev release problem.
1. Definatly announce full releases.
2. Announce and pre- builds like 2.2.0pre9
3. Post about major feature additions.
4. Announce the occasional dev build just to keep people aware of the project.
It would also be more helpful if, instead of saying, here is another dev kernel, people would post about some of the stuff in the kernel, bug-fixes, new feautures, articles about new subsystems, etc. Not everyone is on the kerneldev mailing list you know.
Whoosh, totaly over your head wasn't it? Its funny how /.ers think normal people code video games.
Clue: Video games are coded by corperations, by teams of people. They are infernally attached to custom code. Even though C was available for the Playstation since its introduction, people wrote entire games in ASM for the longest time. Its true that the 32K does not include hardware access libraries, but think about it. What libraries are those? You basically have a set of functions that wrap directly around the hardware. No memory protection so no need to go through the kernel. Developer makes a request, you hit write value x to memory location y and call interupt z, and boom, you're done. The lack of need for a driver model of any sort frees you from half the job right there. Don't expect the PS2 libraries to get much more complex. All it will probably be is a TCP/IP stack. Any services you want to implement will be up to you. And why should it be anything but that? If I can to support fancy stuff I should code it myself and not burden everyone else who just wants the fastest stream of data possible. The only place I can see it getting more complex are some services for Firewire and some TCP/IP services. Not even a network API, just some functions to get a TCP/IP connection running. And I doubt developers want WinCE (wince, get it?) functionality at the cost of speed. Gaming is very competitive and programming more or less directly to the hardware give you a big edge. It might not matter for someone making a browser for the system, but it does to the guy making the game. Developers don't do it to show of their coding skills. They do it because
A) Its easy enough. Only have to write one set of custom code, and no maintainability issues. Once a set of code for a general type of game has been written, then the company re-uses it for future projects. But no need to make a general purpose OS to handle everything, an OS suitable for a driving game would suck when you ran the latest RPG!
B) Its fast as a hell. Crash3 warp would bog down a 133MHz computer with a Voodoo card. Consider this is running on a 33MHz MIPS proc. with a 1994 era graphics chip. It has transparancy, something playstation isn't technically supposed to be able to do. I know you Linux people like abstraction and code reuse and "services at cost of performance" but thats not how its done in gaming. People expect games to continue to get better on the same exact hardware and expect 100% bug free programs.
Alex St. John once said that computer users pay a 80% OS tax on everything and its true. If it really doesn't help, why bother putting an GPOS on the system?
This is almost too funny to be true. Stack up these two machines.
PS2:
-300 MHz processor
-32 meg RAM
-no HD
X-Box
-1000Mhz proc
-64 meg RAM
-4 gig HD
X-Box wins right? Wrong! When you consider that the 300MHz proc in the PS2 is made to run games and only games, you realize that it wipes the floor of even the 1GHz proc. Then you realize, hey! the Windows derived OS will take up half the RAM! Playstation's OS is about 32K and probabaly won't get over 512K in PS2. Okay fine the HD is there. That way MS can issue patches for the system. Then trow in the fact that the PS2 graphics processer whoops the ass of even the GeForce (theoretical 75M/triangles per second vs. theoretical 15M/triangles per second) you see who will trump who. I thing that MS really was really dreamig when they thought this one up.
A. If they can it it in under $300 I will print out the slash code and eat it.
B. If they can get people who are used to one crash in 5 years and bug-free games to live with a crash a week and endless patches, then have the government sue them for using mind control.
MS is in way over its head on this one.
I think there are a few points that will go over most Slashdotter's heads. The first one that I would like to clear up is... DREAMCAST DOES NOT RUN WINCE. A version of it has been made, the web browser uses it I think, but thats about the extent of it. A developer would be stupid to run WinCE, or ANY general purpose OS for that matter, on gaming hardware. Another thing I think a lot of people miss is that this is probably will not run the Windows derived OS for games, just different apps like browsers and such. On a console, the OS is just a bunch of support libraries wrapped around a real time kernel. In the Playstation for example, the OS is just 32K, not just the kernel, the whole thing. Remember, in a console, the system doesn't really need most of the services that other systems need. A network API? Just make a direct interface to TCP/IP. Graphics drivers? Why? Just write some ASM libraries to access graphics hardware directly. Memory protection? Who needs it? Virtual memory, maybe necessary cuz of the HD, but greatly simplified due to the fact that only one process is running. Which brings up another key point. The OS on a console does no governing whatsoever. There is not contention between process for resources. If it is needed, the app accesses it directly. As I remember, most of the first playstation games were written in ASM for the large part. And if you think running Linux on this bad boy is a good idea, all I ask is why? UNIX is a general purpose OS and a very abstracted one at that. Sure it may be cool to get it running, but will it be useful? Ultimatly I think it is a bad idea for general purpose OSs to try to get on game consoles. You're talking about a market where stability is not really key, but people are used to it. I'm not talking UNIX stability either. I have had my N64 for about 5 years. In that time, only one game has frozen on me. One! What is that like a 1600 day uptime? I think you can say that because this machine was being used at 100% for 2 or 3 (not just me, my brother) hours a day on average for 1600 days, easily the load of the linux email server sitting in the corner for 3 years.
You're so wrong I don't even know when to start. Yo, some one moderate this guy down for spreading FUD. I would, but I feel like posting on this thread. First. Win2K is nowhere near 11 years old. Cairo started just before windows 95 came out, around, '94. I know cuz I have the BYTE announcement for it. Cairo is not NT. Cairo was the code name for NT 5.0 which becamse W2K. Second, 95 is not a fork of the 3.1 code base. It was re-written for the most part with most existing code moved into protected mode modules. Its less like the evolution from kernel 2.0 to 2.2 and more like BSD to Next. Next put in a new core and put all the BSD cruft on top of that. Second, NT shares no kernel source with VMS, they don't have the right to. The only VMS stuff NT has is that the designers were originally working on VMS and brought a lot of design (but not code) into NT. I think you have Cario mixed up with NT. Cario is just a version of the OS formarly known as Windows NT, which is itself probably not dated '89. NT was from scratch, but desgin concepts were carried over from VMS. It is a microkernel design which is pretty clean at the core. If you haven't read about the NT kernel, you should, it is pretty nice. Its just the servers on top of the kernel are pieces of shit.
He forgets the counterpoint. If there is only Satan, then the christian goes to hell, and the devil worshippers go to heaven. Similarly if there is only the christian or muslim god (the two religions that believe one who is not one of them will go to hell, then everyone on the wrong side will go to hell. When your talking about something as abstract as theology, you have to be careful because normally logical cases don't apply.