You sound like a person who knows what he's talking about, until you use a phrase like "Phoenix 0.x is 40% faster than IE". That is a meaningless statement and I'm sure you know it. Benchmarking of anything can never be reduced to one ratio.
I use Opera myself, more for the configurability and keyboard control than for the speed, although it rarely makes me wait. I will try Phoenix although I'm not in a hurry. IE is such a piss poor browser that I am continually amazed it is so dominant.
It's true that Perl's basic set of functions expose the Unix API which is implemented in C. But then Perl adds on some additional dynamic sophistication which does not fall within the C execution model. Perl's (bizarre) scoping rules, for example, would require you to write a name-resolver (in C) to follow Perl's rules for name resolution. This goes counter to the normal meaning of compilation in languages like C, in which all names are resolved at compile time. The Perl folks might call it compilation but it doesn't meet my definition of compilation.
It's also a little silly that the language provides no way to specify a single concrete type for the referants of a bunch of different supertype references. For example you might have to write something like this:
List list1 = new java.util.LinkedList(); List list2 = new java.util.LinkedList(); List list3 = new java.util.LinkedList();
You would have to edit the source in three places to make the change to ArrayList. A macro processor would solve the problem. A Pure Java (TM) approach:
List list1 = new java.util.LinkedList(); List list2 = list1.clone(); List list3 = list1.clone();
The only thing an SGI might still have over a PC is that an SGI can do 48bit RGBA, and PC graphics generally dont go past 32bit, and 48bit RGBA is essential for high-fidelity image processing (film work) i've been led to believe. But that will change too - 48bit RGBA cards for PCs will be commonly available within a year or two i'll bet.
The latest greatest PC hardware is the ATI Radeon 9700 card with its DirectX 9 functions. It can accomodate a number of different pixel representations including representations with more than 16 bits per color channel. The representation everyone mentions is 32-bit-per-channel floating point RGBA; 128 bit color. The output to the ramdac is maximum 10-bit per channel, but since your CRT display can't use more than 10-bit color, that doensn't matter. The accuracy of the operations in the graphics memory can far exceed the old Reality Engine accuracy.
I'm not aware of software that actually utilizes high accuracy color so far.
...Perl converted to C? (there is a cmd line option in perl to do this... I have forgotten what it is)...
The only way to translate Perl to C is to create a copy of the Perl interpreter (runtime system) and the Perl compiler, in C, which is what the Perl compiler option that you're thinking of does. The Perl compiler saves you your initial parse and compile stage and that's about it. It does not compile Perl in any ordinary sense of the word. Perl cannot be translated into C, because of its support for such features as runtime execution of dynamically created program text. Other highly dynamic features such as closures would be equally impossible to translate into C without a complete Perl interpreter.
I also recall reading that Germany ignored English copyrights (much to the outrage of English publishing interests) during the rapid development of German industrial strength in the late 1800s/pre WWI period.
Reflective material for traffic signs has pits shaped like the corner of a cube. 90-degree angles, but three of them instead of two. So, any incoming light makes three bounces and exits more or less exactly back in the same direction from which it came.
The same stuff is used for front projection visual effects as first seen in 2001. A half silvered mirror bounces a projected movie onto the scene, on-axis with the camera. The reflective material bounces the movie back into the camera aperture.
My business strategic tool examples are all very real and are in use to make large sums.
1) You can keep innovating faster than the other fellow. That's what any smart company does, including Microsoft. The bulk of their spending is product development & research, that is, designing new stuff. People who don't like what they design might not want to call it innovation, but every year they have a lot of new stuff to show. It's hard to keep up.
2) Monopolies as such are certainly not illegal in the US, although a monopoly holder can't do everything another company can do, and the government may prevent a monopoly from being created through an acquisition. I was referring in particular to legally enforced monopolies such as patents, which were created to permit monopoly pricing.
3) It's true, cartels are illegal here. Too bad that doesn't prevent DeBeers from sucking up American dollars from over in London. An interesting example of the most powerful government on the planet powerless to stop a business strategy.
4) See Nike, Coke, Marlboro, Proctor & Gamble, DeBeers (again): the empires that advertising built. Yes indeed, they do have prices vastly in excess of production costs, due to artificially created demand for products that only those companies have.
Now, if you take a look at the actual market place, you'll see that this is generally true. Most products do not sell that far in excess of production costs. Cars, industrial machines, and even computers, at the large scale, have very low profit margins. Computer manufacturers, for example, make only a few percent profit per machine.
Well, sure, there are lousy businesses out there. Cars, industrial machines and commodity computer assembly probably qualify. (Although opportunities do come along to make crazy margins even on such mature products as these; consider IBM's continuing reign of terror in mainframes, and the US light truck 25% tariff with its resulting 25+% margins on domestic SUVs.) A lousy business is any business that's easy to enter. A good business is one that's impossible or at least very risky and expensive to enter.
Software has always been a good business: it's expensive and risky to do shrink wrapped software development, whether we're talking about office apps, games, operating systems, whatever. Some of this relates to aspects of the product: Switching costs are high so it's relatively easy to get that upgrade revenue. Network effects help protect a monopoly position for some type of software.
But I think the number one reason software is a good business is that it is damn hard to build an effective software development organization. Companies and governments with tremendous resources have tried and failed. Those companies which have the software development chops, like Microsoft, Nintendo, IBM, are able to continue extracting the billions from the rest of us, because they know how to do it and we don't.
If the Germans can fund a successful Exchange/Outlook killer it will be pretty impressive. Lotus/IBM couldn't. Sun couldn't get groupware going despite eons of network cheerleading. We'll see. All I'm saying is, it's far from inevitable.
You wrote: If software just isn't worth the price people are charging for it (and my guess is that the $40 billion in cash in Microsoft's bank indicates the cost of labor is far outstripped by the final selling costs) then eventually something will happen to stabilize the situation again.
If I hear you correctly, you think there is some sort of natural price for all products related to the cost of production, and that the market functions to push the price in that direction. Well it ain't so. Business strategy is, according to one way of looking at it, all about making sure that the price you are selling things for is way higher than the production cost, and keeping it that way. The tools to accomplish this are many: innovation, monopoly by force of law, cartel cooperation, advertising, lock-in, etc. If Germany can overcome Microsoft's marketplace power, good for them, but it's not inevitable.
The PII-233 isn't a system, it's just a cpu+fan which is worth maybe $25 at retail. That's the problem. I don't have a spare Slot-1 motherboard and I'm having a hard time convincing myself it's worthwhile to put a system together around it, as opposed to, say, using a $50 1-GHz Athlon. For now it will remain a spare, since my current system is also Slot-1 running a P3.
I like your idea of not giving admin privileges to friends and relations when I set up a Windows installation. I'd imagine that would increase system stability!
I find it awfully funny how Slashdot UIDs have become of such interest. When they introduced scoring I held out for quite a while before registering. After all, why would I want to register if I could choose not to? Think of all the prestige I gave up by delaying...
AFA the whole CD issue: in the interest of user sanity I would try to work around any requirement that I insert a CD at any time to make a program run. So for example the Windows cab files should be on the HDD so the user doesn't get stuck trying to find the install CD for a network issue. Games should be patched. If there's no other way you can work around it by faking the CD with Daemon Tools. Users seem to be really bad at finding CDs when they need them. And I do not exclude myself from that judgement.
Thanks for the reply. I admit I have seen the phenomenon you're talking about: regular users, when their machine is fubar'd, will throw it away and buy a new one, unless they have a computer geek at their disposal to make it nice again. And they do often give up after bringing home software that doesn't work well, rather than trying to analyze and fix the problem. Which usually has something to do with a ton of running software that they don't need.
Meanwhile I got to win my trivial point about four year old machines and RTCW. Happy slashdotters all around.
Should users install applications? I agree with you; they should not. I think their machines should be professionally managed as a service, and normally the user's data should not even be in the home. But that's all a long way off. For now the closest approximation is something like WebTV + Hotmail + Yahoo Games, something too low-functioning to recommend to most anyone. For today there is no practical alternative to purchasing "Backgammon Madness" at CompUSA, ramming the CD in the drive, and letting AutoPlay work its magic. It's a pity that Linux doesn't make that realistic for Joe Sixpack.
I like old hardware too. I'm holding on to a PII-233, just in case it ever comes in handy. As a paperweight, I suppose. I was using a four-year-old 3D Labs VX1 video card until someone just randomly gave me an 18-month-year-old ATI Radeon card. Yay, now I can try RTCW too.
9M pixels at 32bpp is 36M for a full screen framebuffer. Cards with >= 128M onboard memory are commonplace and there's nowhere to go but up. So the amount of memory isn't a big problem.
Moving a 4M pixel window in real time isn't a problem either. Actually that kind of operation is so parallelizable that the peak performance of today's hardware could be brought to bear. That's about 2.4 billion pixel per second fill on a Radeon 9700. There's enough headroom there to move a 40M pixel window at a high frame rate.
Glad to know you, Corporate, or do you prefer Mr. Troll? The first point is apparently a difference of opinion; I would certainly say that "Dell and Compaq should not sell unproven hardware with flaky drivers". When a component is not ready to be a part of a functional complete system, I think the integrators should stay away for the sake of their continuing business with customers as well as for the sake of minimizing support costs. On to the second point.
Your remarks about users not knowing their requirements, and about poor maintenance, are well taken. But what I said was: you can run RTCW on a four year old machine. Let's look at this.
The computer I mentioned (the PPro 200, 256Meg RAM, with Voodoo2) is able to run RTCW but you have to set detail very very low.
That is not really a good machine from four years ago. The Pentium Pro was introduced in October 1995; the PII in April 1997. Four years ago in September 1998, a hot gaming machine was a P2-400 with 128M and a 16MB TNT, Voodoo2 or perhaps even an SLI setup, running Win98. Today, the top three selling video games from July 2002 were Warcraft 3, the Sims, and a Sims expansion pack. The official requirements for these games are all within this range:
Warcraft 3 Minimum Requirements 400MHz Processor, 128MB RAM, 8MB Video Card, 700MB HDD Space
The Sims Minimum Requirements P233 MHz, 32 MB RAM, 2 MB Video Card, 4x CD-ROM, 300MB hard drive space
RTCW is in the same class: Minimum Requirements PII 400, 128MB RAM, 16MB 3D acclerated video card, 800MB hard disk space, 4x CD-ROM
As to Neverwinter Nights (number four on the July list) its requirements are higher: Minimum Requirements PII 450 MHz, 96MB RAM, 16MB TNT2-class video card
You would need a good PC from three and a half years ago to play that one.
The reason all these games run on such old computers is that people are still using their old computers! The marketing department won't let you release a product that curls up and dies on a good, four-year-old machine.
To be sure, if you bought a crappo discount PC four years ago, you would need to spend probably $100 on RAM and video upgrades to make it good enough to play today's games. (I recently performed such surgery on my girlfriend's crappy four year old PC, adding a $30 3D video card that made it capable of playing Quake 2 for the first time in its life.) But then, a crappo discount PC won't play current games well at the time of its manufacture, either. As you say: users need to define their requirements.
Getting back to the original point, I would continue to claim that a four year old (decent) PC is still ready to run today's popular software. And just to relate this to the original "girlfriend buys You Don't Know Jack" scenario:
You Don't Know Jack 5th Dementia Minimum Requirements P200, 64MB RAM, 8x CD-ROM, 275MB hard drive space
I doubt you could find a PC on shelves four years ago that didn't meet the requirements to install and run You Don't Know Jack. So I'm with nutbar: Windows does a better job with enabling users to buy third party software, and slightly old hardware is not too much of a problem.
Hey, wait, you're that AC troll who keeps contradicting me in the vaguest possible terms. I'm not showing you my authority! You show me your authority to question my authority first! Then we'll talk.
That is the job of the integrator: to sell a system that works well. That means waiting until the drivers are good enough before a component is integrated into the system. You're correct that most folks will never upgrade drivers, but they shouldn't have to.
As to the performance problems: this is overstated. The user base for performance intensive applications like RTCW is small, for one thing. And for another thing, you can run RTCW on a four year old machine. It just runs slow. If it can't fit in main memory, perhaps very slow. But my point is that backwards hardware compatibility for new APIs is pretty darn good in the Windows world. You just need to add a stick of memory now and again.
That is absolutely correct. The sort of "regular person" who would never install Linux, would also never install Windows. They would rather pay the $1000 for a new computer than attempt any hardware or O/S surgery. The "rescue cd" might be employed from time to time; it is actually a very helpful tool for the completely clueless.
From time to time they might try a Windows upgrade. This usually has a tragic end.
Re:With the unix penchant for 2 letter abbrevs
on
Vi IMproved -- Vim
·
· Score: 1
Actually, Orthonormal, it may interest you that "vi" is an abbreviation for "visual". In fact vi/ex accept abbreviations for commands. So "vi" is what you type to enter visual (vi) mode from ex, but "vis", "visu", "visua" and "visual" are also accepted.
ITS was built not in Lisp but assembly. That's why ITS isn't available on current hardware. Ref. Steven Levy's _Hackers_. It is, however, true that versions of emacs were the standard editors on many different Lisp based systems.
On a modem Opera is tremendously faster than IE if used properly. Image loading should be turned to "Show cached images only". Links should be opened in the background when possible: "shift-ctrl-click". Images should be loaded only when necessary: "shift-g" for one page, or "g" to change the image loading for a window if there's a lot of graphical text etc. Popups should normally be left off.
With Opera configured this way, most web sites are twice as fast as on a regular browser that loads images. Navigation images can be loaded just once if desired and will be displayed from then on. (Opera seems to cache more effectively than IE.) And with background loading you can easily avoid ever sitting and waiting for the browser.
IE can toggle image display but only by going through a graphical dialog. There's nothing like Opera's "cached only" which is usually just right. And though IE does have window spawning, it doesn't work as well as Opera's. For example there's no such thing as background spawning; spawned windows do not open full screen; and there's no SDI option, which I prefer.
Yup: I'm another one of those walking contradictions who downloaded and watched AOTC and then proceeded to go down to the movie theater and buy a ticket. Makes ya think about all that "lost revenue".
Slashdot Mirror
You sound like a person who knows what he's talking about, until you use a phrase like "Phoenix 0.x is 40% faster than IE". That is a meaningless statement and I'm sure you know it. Benchmarking of anything can never be reduced to one ratio.
I use Opera myself, more for the configurability and keyboard control than for the speed, although it rarely makes me wait. I will try Phoenix although I'm not in a hurry. IE is such a piss poor browser that I am continually amazed it is so dominant.
It's true that Perl's basic set of functions expose the Unix API which is implemented in C. But then Perl adds on some additional dynamic sophistication which does not fall within the C execution model. Perl's (bizarre) scoping rules, for example, would require you to write a name-resolver (in C) to follow Perl's rules for name resolution. This goes counter to the normal meaning of compilation in languages like C, in which all names are resolved at compile time. The Perl folks might call it compilation but it doesn't meet my definition of compilation.
The only thing an SGI might still have over a PC is that an SGI can do 48bit RGBA, and PC graphics generally dont go past 32bit, and 48bit RGBA is essential for high-fidelity image processing (film work) i've been led to believe. But that will change too - 48bit RGBA cards for PCs will be commonly available within a year or two i'll bet.
The latest greatest PC hardware is the ATI Radeon 9700 card with its DirectX 9 functions. It can accomodate a number of different pixel representations including representations with more than 16 bits per color channel. The representation everyone mentions is 32-bit-per-channel floating point RGBA; 128 bit color. The output to the ramdac is maximum 10-bit per channel, but since your CRT display can't use more than 10-bit color, that doensn't matter. The accuracy of the operations in the graphics memory can far exceed the old Reality Engine accuracy.
I'm not aware of software that actually utilizes high accuracy color so far.
...Perl converted to C? (there is a cmd line option in perl to do this... I have forgotten what it is)...
The only way to translate Perl to C is to create a copy of the Perl interpreter (runtime system) and the Perl compiler, in C, which is what the Perl compiler option that you're thinking of does. The Perl compiler saves you your initial parse and compile stage and that's about it. It does not compile Perl in any ordinary sense of the word. Perl cannot be translated into C, because of its support for such features as runtime execution of dynamically created program text. Other highly dynamic features such as closures would be equally impossible to translate into C without a complete Perl interpreter.
Perhaps you are thinking of oval bicycle chainrings?
"Biopace is a patented non-round chainwheel design made and licensed by Shimano."
I also recall reading that Germany ignored English copyrights (much to the outrage of English publishing interests) during the rapid development of German industrial strength in the late 1800s/pre WWI period.
Reflective material for traffic signs has pits shaped like the corner of a cube. 90-degree angles, but three of them instead of two. So, any incoming light makes three bounces and exits more or less exactly back in the same direction from which it came.
The same stuff is used for front projection visual effects as first seen in 2001. A half silvered mirror bounces a projected movie onto the scene, on-axis with the camera. The reflective material bounces the movie back into the camera aperture.
Hopefully they'll reprint it with a real cover. I have the crappy Yourdon Press edition.
My business strategic tool examples are all very real and are in use to make large sums.
1) You can keep innovating faster than the other fellow. That's what any smart company does, including Microsoft. The bulk of their spending is product development & research, that is, designing new stuff. People who don't like what they design might not want to call it innovation, but every year they have a lot of new stuff to show. It's hard to keep up.
2) Monopolies as such are certainly not illegal in the US, although a monopoly holder can't do everything another company can do, and the government may prevent a monopoly from being created through an acquisition. I was referring in particular to legally enforced monopolies such as patents, which were created to permit monopoly pricing.
3) It's true, cartels are illegal here. Too bad that doesn't prevent DeBeers from sucking up American dollars from over in London. An interesting example of the most powerful government on the planet powerless to stop a business strategy.
4) See Nike, Coke, Marlboro, Proctor & Gamble, DeBeers (again): the empires that advertising built. Yes indeed, they do have prices vastly in excess of production costs, due to artificially created demand for products that only those companies have.
Now, if you take a look at the actual market place, you'll see that this is generally true. Most products do not sell that far in excess of production costs. Cars, industrial machines, and even computers, at the large scale, have very low profit margins. Computer manufacturers, for example, make only a few percent profit per machine.
Well, sure, there are lousy businesses out there. Cars, industrial machines and commodity computer assembly probably qualify. (Although opportunities do come along to make crazy margins even on such mature products as these; consider IBM's continuing reign of terror in mainframes, and the US light truck 25% tariff with its resulting 25+% margins on domestic SUVs.) A lousy business is any business that's easy to enter. A good business is one that's impossible or at least very risky and expensive to enter.
Software has always been a good business: it's expensive and risky to do shrink wrapped software development, whether we're talking about office apps, games, operating systems, whatever. Some of this relates to aspects of the product: Switching costs are high so it's relatively easy to get that upgrade revenue. Network effects help protect a monopoly position for some type of software.
But I think the number one reason software is a good business is that it is damn hard to build an effective software development organization. Companies and governments with tremendous resources have tried and failed. Those companies which have the software development chops, like Microsoft, Nintendo, IBM, are able to continue extracting the billions from the rest of us, because they know how to do it and we don't.
If the Germans can fund a successful Exchange/Outlook killer it will be pretty impressive. Lotus/IBM couldn't. Sun couldn't get groupware going despite eons of network cheerleading. We'll see. All I'm saying is, it's far from inevitable.
You wrote: If software just isn't worth the price people are charging for it (and my guess is that the $40 billion in cash in Microsoft's bank indicates the cost of labor is far outstripped by the final selling costs) then eventually something will happen to stabilize the situation again.
If I hear you correctly, you think there is some sort of natural price for all products related to the cost of production, and that the market functions to push the price in that direction. Well it ain't so. Business strategy is, according to one way of looking at it, all about making sure that the price you are selling things for is way higher than the production cost, and keeping it that way. The tools to accomplish this are many: innovation, monopoly by force of law, cartel cooperation, advertising, lock-in, etc. If Germany can overcome Microsoft's marketplace power, good for them, but it's not inevitable.
The PII-233 isn't a system, it's just a cpu+fan which is worth maybe $25 at retail. That's the problem. I don't have a spare Slot-1 motherboard and I'm having a hard time convincing myself it's worthwhile to put a system together around it, as opposed to, say, using a $50 1-GHz Athlon. For now it will remain a spare, since my current system is also Slot-1 running a P3.
I like your idea of not giving admin privileges to friends and relations when I set up a Windows installation. I'd imagine that would increase system stability!
I find it awfully funny how Slashdot UIDs have become of such interest. When they introduced scoring I held out for quite a while before registering. After all, why would I want to register if I could choose not to? Think of all the prestige I gave up by delaying...
AFA the whole CD issue: in the interest of user sanity I would try to work around any requirement that I insert a CD at any time to make a program run. So for example the Windows cab files should be on the HDD so the user doesn't get stuck trying to find the install CD for a network issue. Games should be patched. If there's no other way you can work around it by faking the CD with Daemon Tools. Users seem to be really bad at finding CDs when they need them. And I do not exclude myself from that judgement.
Wicked Problems, Righteous Solutions [amazon.com] , written in 1998
1990 actually.
Thanks for the reply. I admit I have seen the phenomenon you're talking about: regular users, when their machine is fubar'd, will throw it away and buy a new one, unless they have a computer geek at their disposal to make it nice again. And they do often give up after bringing home software that doesn't work well, rather than trying to analyze and fix the problem. Which usually has something to do with a ton of running software that they don't need.
Meanwhile I got to win my trivial point about four year old machines and RTCW. Happy slashdotters all around.
Should users install applications? I agree with you; they should not. I think their machines should be professionally managed as a service, and normally the user's data should not even be in the home. But that's all a long way off. For now the closest approximation is something like WebTV + Hotmail + Yahoo Games, something too low-functioning to recommend to most anyone. For today there is no practical alternative to purchasing "Backgammon Madness" at CompUSA, ramming the CD in the drive, and letting AutoPlay work its magic. It's a pity that Linux doesn't make that realistic for Joe Sixpack.
I like old hardware too. I'm holding on to a PII-233, just in case it ever comes in handy. As a paperweight, I suppose. I was using a four-year-old 3D Labs VX1 video card until someone just randomly gave me an 18-month-year-old ATI Radeon card. Yay, now I can try RTCW too.
9M pixels at 32bpp is 36M for a full screen framebuffer. Cards with >= 128M onboard memory are commonplace and there's nowhere to go but up. So the amount of memory isn't a big problem.
Moving a 4M pixel window in real time isn't a problem either. Actually that kind of operation is so parallelizable that the peak performance of today's hardware could be brought to bear. That's about 2.4 billion pixel per second fill on a Radeon 9700. There's enough headroom there to move a 40M pixel window at a high frame rate.
Glad to know you, Corporate, or do you prefer Mr. Troll? The first point is apparently a difference of opinion; I would certainly say that "Dell and Compaq should not sell unproven hardware with flaky drivers". When a component is not ready to be a part of a functional complete system, I think the integrators should stay away for the sake of their continuing business with customers as well as for the sake of minimizing support costs. On to the second point.
Your remarks about users not knowing their requirements, and about poor maintenance, are well taken. But what I said was: you can run RTCW on a four year old machine. Let's look at this.
The computer I mentioned (the PPro 200, 256Meg RAM, with Voodoo2) is able to run RTCW but you have to set detail very very low.
That is not really a good machine from four years ago. The Pentium Pro was introduced in October 1995; the PII in April 1997. Four years ago in September 1998, a hot gaming machine was a P2-400 with 128M and a 16MB TNT, Voodoo2 or perhaps even an SLI setup, running Win98. Today, the top three selling video games from July 2002 were Warcraft 3, the Sims, and a Sims expansion pack. The official requirements for these games are all within this range:
Warcraft 3 Minimum Requirements
400MHz Processor, 128MB RAM, 8MB Video Card, 700MB HDD Space
The Sims Minimum Requirements
P233 MHz, 32 MB RAM, 2 MB Video Card, 4x CD-ROM, 300MB hard drive space
RTCW is in the same class:
Minimum Requirements
PII 400, 128MB RAM, 16MB 3D acclerated video card, 800MB hard disk space, 4x CD-ROM
As to Neverwinter Nights (number four on the July list) its requirements are higher:
Minimum Requirements
PII 450 MHz, 96MB RAM, 16MB TNT2-class video card
You would need a good PC from three and a half years ago to play that one.
The reason all these games run on such old computers is that people are still using their old computers! The marketing department won't let you release a product that curls up and dies on a good, four-year-old machine.
To be sure, if you bought a crappo discount PC four years ago, you would need to spend probably $100 on RAM and video upgrades to make it good enough to play today's games. (I recently performed such surgery on my girlfriend's crappy four year old PC, adding a $30 3D video card that made it capable of playing Quake 2 for the first time in its life.) But then, a crappo discount PC won't play current games well at the time of its manufacture, either. As you say: users need to define their requirements.
Getting back to the original point, I would continue to claim that a four year old (decent) PC is still ready to run today's popular software. And just to relate this to the original "girlfriend buys You Don't Know Jack" scenario:
You Don't Know Jack 5th Dementia Minimum Requirements
P200, 64MB RAM, 8x CD-ROM, 275MB hard drive space
I doubt you could find a PC on shelves four years ago that didn't meet the requirements to install and run You Don't Know Jack. So I'm with nutbar: Windows does a better job with enabling users to buy third party software, and slightly old hardware is not too much of a problem.
Hey, wait, you're that AC troll who keeps contradicting me in the vaguest possible terms. I'm not showing you my authority! You show me your authority to question my authority first! Then we'll talk.
I recommend that your four-year-old switch to NTFS for any NT partitions. Her disk corruption troubles will be over very quickly.
That is the job of the integrator: to sell a system that works well. That means waiting until the drivers are good enough before a component is integrated into the system. You're correct that most folks will never upgrade drivers, but they shouldn't have to.
As to the performance problems: this is overstated. The user base for performance intensive applications like RTCW is small, for one thing. And for another thing, you can run RTCW on a four year old machine. It just runs slow. If it can't fit in main memory, perhaps very slow. But my point is that backwards hardware compatibility for new APIs is pretty darn good in the Windows world. You just need to add a stick of memory now and again.
That is absolutely correct. The sort of "regular person" who would never install Linux, would also never install Windows. They would rather pay the $1000 for a new computer than attempt any hardware or O/S surgery. The "rescue cd" might be employed from time to time; it is actually a very helpful tool for the completely clueless.
From time to time they might try a Windows upgrade. This usually has a tragic end.
Your question is inadequate; please restate.
Actually, Orthonormal, it may interest you that "vi" is an abbreviation for "visual". In fact vi/ex accept abbreviations for commands. So "vi" is what you type to enter visual (vi) mode from ex, but "vis", "visu", "visua" and "visual" are also accepted.
ITS was built not in Lisp but assembly. That's why ITS isn't available on current hardware. Ref. Steven Levy's _Hackers_. It is, however, true that versions of emacs were the standard editors on many different Lisp based systems.
On a modem Opera is tremendously faster than IE if used properly. Image loading should be turned to "Show cached images only". Links should be opened in the background when possible: "shift-ctrl-click". Images should be loaded only when necessary: "shift-g" for one page, or "g" to change the image loading for a window if there's a lot of graphical text etc. Popups should normally be left off.
With Opera configured this way, most web sites are twice as fast as on a regular browser that loads images. Navigation images can be loaded just once if desired and will be displayed from then on. (Opera seems to cache more effectively than IE.) And with background loading you can easily avoid ever sitting and waiting for the browser.
IE can toggle image display but only by going through a graphical dialog. There's nothing like Opera's "cached only" which is usually just right. And though IE does have window spawning, it doesn't work as well as Opera's. For example there's no such thing as background spawning; spawned windows do not open full screen; and there's no SDI option, which I prefer.
Yup: I'm another one of those walking contradictions who downloaded and watched AOTC and then proceeded to go down to the movie theater and buy a ticket. Makes ya think about all that "lost revenue".