I still think that it does a decent job of getting the idea across, though, since most people will assume the stereotypical firefighter, which isn't a firebug.
That's odd. Thanks to the Internet, it took all of a minute to download a copy of "A Painted House" and grep through it to check the attribution -- I can't find any instances of "fireman" in the entire text. None of the sentences that contained "soldier" looked anything like this quote.
Can you provide the actual quote from "A Painted House" that you're thinking of? Uggy specifically claimed that this quote was original and from him. I *would* like to give credit where credit is due, but I honestly can't find what you're thinking of.
Go to hotbot.com, click on "Advanced Search", check the "MS Word" box under "Page Content". Then search for whatever you're interested in.
For example, checking this box and then searching for "sco" returns 4600 web pages containing a link to a.doc file relating to SCO. One wonders what facinating goodies might be hidden in metadata in SCO documents...
But note: if the goal is to "legitimize" p2p so that artists get paid, how would you do it?
P2P is already legitimate. P2P has never been illegitimate. The statement has as much basis as "knives are illegitimate", "fire is illegitimate" or "sports cars are illegitimate".
The problem is that if wealth has a direct impact on the operations of government, it's easy to form a feedback loop where the richer just get more rich and powerful, and the entire nation moves further and further away from the (ideal) meritocracy that benefits almost everyone in the long run.
As a P2P software developer and distributor, we believe you have the ability and responsibility to better educate consumers about these known risks, and to design your software in a manner that minimizes the risks. We view with grave concern reports that at least some P2P software developers may be adding features deliberately designed to hinder law enforcement in its prosecution of crimes using P2P software. Companies that engage in such conduct, and fail to meet the important responsibilities referenced above, harm the interests of consumers in our States.
Yes. God forbid we have anonymity or encryption.
[shrug] Well, as I said earlier, I have no interest in following directives like these. Software can be developed privately and via anonymous access through Freenet if necessary. It'd be a pain in the ass, but I'm
* Not interested in adding back doors to my work
* Not interested in stopping work on problems of how to provide secure/nonabusable/anonymous P2P systems (yes, part of that is to benefit users concerned about law enforcement attention).
If the AG wants to do something to go after people operating in legal gray area, he can go after people with radar detectors (speeding can, y'know, kill people, whereas a pirated song only means that a large company gets a small amount less money), or those committing corporate accounting hanky-panky, or any number of other more damaging actions. Admittedly, there aren't people with deep pockets and old-boy connections to the government trying to finance hunting people down (note: AG can also go after corrupt government officials, IMHO), but theoretically that AG was appointed to be the servant of the people, and as the House is demonstrating, popular support for the RIAA is awfully low.
Okay, "reviewed" may be a legitimate complaint (though I find it disturbing that flow-of-control of legislation would go through the MPAA, rather than sending him a *copy*).
However, "drafted" is roughly equivalent to "authored", for our purposes. We don't care whether the guy could spell -- we care about who is coming up with the bulk of the ideas.
I couldn't find any Slashdot articles with comments (?) from '98 when I did a search for "GNOME". GNOME 1.0 was released in '99, and was the first Slashdot story with many comments on GNOME, so I looked in there.
Right up near the top, I get "Gnome is for sysadmins - not home users" as a comment title. Searching for "Windows" and "Microsoft" fails to bring up inflammatory comments. I scrolled down, looking at article titles, and saw "not there yet", "Who cares about Joe Sixpack", and "fvwm is the best". I skimmed through the entire first page (I'm not going to do all the pages) and I couldn't see a *single* comment that could be easily identified by its title as being someone claiming that Windows was in trouble or Linux was going to be big on the desktop. Lots of talk about goals, lots of GNOME/KDE bashing (at least some things stay the same), no talking about "this being the year". A lot of people saying that the release was premature.
I think that you're wrong. Most of the talk about Linux was based on the groundswell of support for Linux on servers. There was a lot of "Try Linux! You can use it instead of Windows now!" starting, IIRC, sometime around 2001, but that was on Slashdot and aimed at techies".
KDE libs might have been out in 98, but there was no set of applications to replace Windows apps out.
When people say "this year Linux is going to be big on the desktop", they may be optimistic, but they aren't way out in left field, doing a five-year-early-prediction.
I can't understand what your proposal is, unless you have a very vague concept of where you're going. How are you suggesting one interface with the compiler, describe your requirements? My issue is that it seems that no matter what you do, you're going to be bringing some set of constraints (potentially incorrect) into the ballgame.
As for provability -- CMU has research on proof-carrying Java code, where there are certain things proved about the code. I can also tell you that it is not possible to prove arbitrary things about any piece of code, or you'd have solved the halting problem -- "Will this program get stuck or not?"
I can give examples of how they *might* do it. I didn't RTFA either.
* Pointer arithmetic
Pointer arithmetic becomes an operation that calls a function that knows what chunks of memory were allocated and checks before mucking with it.
* Hardcoded type sizes instead of using sizeof() (i.e. assuming sizeof(int) == 4).
I don't see why there would be any problem doing this. This isn't correct C code, though, and hasn't ever been (that being said, C does not natively provide certain guarantees that people require for basic functionality, so I suspect most C code is somewhat broken WRT integer sizes).
* Lax rules for casting
Why would there be a problem? If pointer dereferences are checked, what's wrong with misinterpreting a float as an integer or visa versa?
You know, while I *vastly* prefer to use C-based applications over Java-based ones, I have to say that your example is one of the ones where I would consider Java before C.
You have an application that is presumably a custom job. Developer time is expensive and resource usage is cheap on custom or vertical market jobs. Furthermore, Java does a better job of providing built-in safety checks, and your application deals with money.
If you want programs to dump stack traces, it's pretty easy. You can modify your shell set up to dump out stack traces when a program dies.
Windows 98 has protected memory. You can't do very effective virtual memory in the modern sense of the word without implementing most of what you need to have protected memory.
The main problem that you have, I think, is that a sandbox is a high-overhead system that does fancier error output, handling dynamic typing and the like, killing a program the moment it steps a byte out of line. Since hardware solutions like the MMU aren't *quite* this smart, you can't kill things exactly when they start to behave naughtily.
It would be best if we could move on from procedural/imperative/object-oriented/functional languages to a language based on logic that can be used to prove program goals.
I've thought about this, and pretty much decided that there isn't a really feasible way of doing this. Setting a bunch of assertions and letting the compiler "do the right thing" is, I think, an attractive but ephemeral dream. Why?
Think about what you'd do when, say, sorting a linked list. In a traditional language, you'd shove some pointers around. In a constraint-based language, you'd say something like "after this point, I want this assertion to hold -- for all nodes N1, N2 such that N2=N1->next->next...->next, N2->data <= N1-data". The compiler figures out some sort of efficient way to do up the code. The problem is, you've already imposed a data structure on the compiler, which might not be the right one. What if you want to remove an element from the list? You can do "after this point, I want this assertion to hold -- for all nodes N1 such that N1=head->next->next...>next, N1->data != 3; Then you realize that you just removed an employee from the list before assigning him his final paycheck -- you made a higher-level logic error.
Short of producing a full-blown humanlike AI that guesses and lets you refine, where you can say "Give me a word processor", "Okay, add a spellchecker to it", "Okay, make the menu item to invoke the spellchecker be under the Edit menu from under the File menu, where you dumped it", I don't think it's really feasible to go for really high-level languages -- you just don't get the benefits you think you do.
I've taken a look at aspect-oriented programming, and I've come to the conclusion that it will not catch on all over unless the current programming model radically changes.
The problem with aspect-oriented programming is that it can be very difficult to determine where code is located in your source tree that is causing a particular effect that you're observing at runtime. With a procedural language (and barring extensive use of, say, function pointers), it's generally a fairly straightforward task to locate what code is being run that's causing something to break when you perform operation X.
This can admittedly be solved -- as long as one uses an intelligent IDE that can tell you what code affects what code at runtime when you're poking through source.
The problem is that doing this requires everyone adopting a particular IDE (maybe an extended Eclipse) for effective use.
The programming model that dates back to the earliest compilable programming systems is that you produce text in an ordinary text-editing program, and then compile it. You don't need a special program to work with your project.
It doesn't mean that such a move can't be made, but I do think that AOP will not catch on in general until a move like everyone starting to use Eclipse for software development happens.
If you are writing applications its better to use Java or C# anyday.
You know, I've heard that, and I don't agree, at least as a general rule.
For custom code development or vertical market stuff, it's reasonable, unless there are other significant concerns.
However, an end-product written in C is faster and more memory-efficient. If someone proposed rewriting bash in Java, I'd never use what they put out. Same for the GIMP, xterm, xmms, or any of the other apps that dot my desktop.
Of the Java apps I've tried...I think I've tried straw (RSS reader, found it to be slow and ended up not using it), tube (Java Hotline client, ended up using the C-based fidelio in preference), freenet (still no good C alternative, so I don't run freenet, as it soaks up masses of RAM), and a bunch of other minor apps that just got quickly dropped. If you're writing something that many people are going to be using all over the world, and you want a really polished final product, I claim that you're generally better off with C (and I wish that I could think of a good safe language, but the only safe languages that I've been reasonably impressed with for writing full-blown applications -- eiffel and ocaml -- have a stunning lack of interest from the public).
* C is mature and stable. Yup, there are some bad design decisions, but using C isn't quite like riding the current version of Java -- new idea, whoops, let's change that, oh wait, let's extend that...
* There are good development tools for C. There is a good set of tools that have been built up over the years for analyzing and debugging C.
* C doesn't impose a lot of resource overhead. I'm all for having safe languages, but Java is just plain fat when it comes to memory usage, and encourages a lot of practices (like rapid destruction and creation of objects) that tends to cause a serious CPU hit as well. Safe and *efficient* languages have, as a general rule, flopped in the marketplace.
* C runs on everything. There are awfully few platforms for which there isn't a C compiler.
There are a lot of Java apps out there. Yet, in my day-to-day use of my computer, I don't use a single Java-based program. Not one. I use a small number of perl and python scripts, a couple of bash scripts, and a vast quantity of C code. The C alternatives are faster and more memory-efficient. Plus, I've tended to come to believe that the average Java programmer experience level is lower than than of the average C programmer (simply because lots of Java people *started* coding with Java which puts an upper bound on experience).
I admit that Java can be neat for some things (rapid application development/prototyping, lightweight distributed and network-using apps). It's generally not the sort of thing that sits on my desktop, though.
It may be that more Java code is written each day now than C code, but since so much Java development is vertical market or custom development, I suspect that the overwhelming majority of lines of code *used* by people is C.
Okay, I'm not a DB programmer, so I just may be out of touch here, but I know that Postgres had transactions, I'm quite sure that Oracle had transactions, and I strongly suspect that MS SQL had transaction support already.
If MySQL really did just get transaction support, it seems more like My was a bit behind, rather than being far ahead of the others.
Well, perhaps nobody else does, but I buy my video cards based on open source support (not Linux support alone).
For years, Matrox had the best support for Linux with open source drivers. I bought Matrox cards. Currently, ATI has the best open source support. Right now, I'm buying ATI. I'll keep doing so, as well. I use my cards under Linux exclusively, and binary drivers are a tremendous pain in the ass to deal with. I recognize that video card vendors have reasons for wanting to keep their drivers closed-source -- that's fine, but I happen to value open source.
Depends on how you chose to set it up (or in the case of Joe Six-pack, how the packager does). It's not that hard to grant ordinary users write privs to the CD device.
You're full of it. You are trying to say, with a straight face, that people expected Linux to take over the desktop in *'98*?
Most of the time in the past was people getting excited about the ability for *geeks* to use exclusively Linux -- Open Office, Samba, etc made it feasible to work with Windows users and still keep using Linux.
Red Hat's CEO said, what, six months ago that Linux isn't ready for the desktop war just yet?
This year and last year are big because there are a lot of major open source apps coming out and being *usable*, by *typical users*, at at least a basic level, as a substitute for Windows apps.
Finally, if you don't think Linux usability has improved massively since '98, you just plain don't remember 98. We had no GNOME or KDE apps. Preference dialogs didn't exist. Widget sets were Tk, and black-and-white Athena. Boxes required a serious sysadmin to secure out-of-box.
Last year, I agree that there were a lot of people on Slashdot that were predicting big gains on the desktop. And guess what? A bunch of governments and big companies starting transition processes, or at least made it much more easy to move a chunk at a time to Linux. If anything, I'm surprised that things are going this quickly.
My prediction is that Linux will break 10% desktop market share before the end of 2006. That is a *huge* number of users to move from one platform to another -- perhaps around 100 million users -- , but remember that there's a threshhold effect at which point application vendors, people doing file formats, etc cannot ignore Linux, and once that hump is over, it becomes much easier to move to Linux.
Web sites are already improving -- I don't see the number of "IE-only" sites that I did thanks to the spread of Mozilla, Linux, and Mac OS X running Safari.
That being said, I think that as Microsoft gets more worried, they will do whatever it takes to fight back effectively. That may be as far as moving to a Linux-based distribution and porting their products to it. Microsoft is unlikely to die, no matter what.
I don't believe that 20% of all corporate desktops in the world will be running Linux by the end of 2005. No, that just isn't plausible.
Businesses do transitions when they bring in new machines. A typical business machine has, what, a four year lifecycle? That means that even if every machine being replaced was converted to Linux, there'd be less than 50% people using Linux. You're requiring about half of all new desktop purchases to be Linux-based. That's wildly unrealistic.
I'd say that 20% Linux desktop business penetration by the end of 2007 would be a very positive outcome.
Slashdot Mirror
Shakespeare and artists of the time had their work duplicated all over.
The question is whether the IP laws we put into place have produced a worthwhile increase in content production in the arts.
My personal take is "sometimes yes, sometimes no".
Yeah, Uggy said the same thing. :-)
I still think that it does a decent job of getting the idea across, though, since most people will assume the stereotypical firefighter, which isn't a firebug.
Ah, we're a cynical bunch.
That's odd. Thanks to the Internet, it took all of a minute to download a copy of "A Painted House" and grep through it to check the attribution -- I can't find any instances of "fireman" in the entire text. None of the sentences that contained "soldier" looked anything like this quote.
Can you provide the actual quote from "A Painted House" that you're thinking of? Uggy specifically claimed that this quote was original and from him. I *would* like to give credit where credit is due, but I honestly can't find what you're thinking of.
Damn, and I always loved that free candy that they bundled with my tennis shoes. Well, so much for shoe brand loyalty for me!
Go to hotbot.com, click on "Advanced Search", check the "MS Word" box under "Page Content". Then search for whatever you're interested in.
.doc file relating to SCO. One wonders what facinating goodies might be hidden in metadata in SCO documents...
For example, checking this box and then searching for "sco" returns 4600 web pages containing a link to a
But note: if the goal is to "legitimize" p2p so that artists get paid, how would you do it?
P2P is already legitimate. P2P has never been illegitimate. The statement has as much basis as "knives are illegitimate", "fire is illegitimate" or "sports cars are illegitimate".
The problem is people trying to *il*legalize it.
The problem is that if wealth has a direct impact on the operations of government, it's easy to form a feedback loop where the richer just get more rich and powerful, and the entire nation moves further and further away from the (ideal) meritocracy that benefits almost everyone in the long run.
As a P2P software developer and distributor, we believe you have the ability and responsibility to better educate consumers about these known risks, and to design your software in a manner that minimizes the risks. We view with grave concern reports that at least some P2P software developers may be adding features deliberately designed to hinder law enforcement in its prosecution of crimes using P2P software. Companies that engage in such conduct, and fail to meet the important responsibilities referenced above, harm the interests of consumers in our States.
Yes. God forbid we have anonymity or encryption.
[shrug] Well, as I said earlier, I have no interest in following directives like these. Software can be developed privately and via anonymous access through Freenet if necessary. It'd be a pain in the ass, but I'm
* Not interested in adding back doors to my work
* Not interested in stopping work on problems of how to provide secure/nonabusable/anonymous P2P systems (yes, part of that is to benefit users concerned about law enforcement attention).
If the AG wants to do something to go after people operating in legal gray area, he can go after people with radar detectors (speeding can, y'know, kill people, whereas a pirated song only means that a large company gets a small amount less money), or those committing corporate accounting hanky-panky, or any number of other more damaging actions. Admittedly, there aren't people with deep pockets and old-boy connections to the government trying to finance hunting people down (note: AG can also go after corrupt government officials, IMHO), but theoretically that AG was appointed to be the servant of the people, and as the House is demonstrating, popular support for the RIAA is awfully low.
Okay, "reviewed" may be a legitimate complaint (though I find it disturbing that flow-of-control of legislation would go through the MPAA, rather than sending him a *copy*).
However, "drafted" is roughly equivalent to "authored", for our purposes. We don't care whether the guy could spell -- we care about who is coming up with the bulk of the ideas.
I couldn't find any Slashdot articles with comments (?) from '98 when I did a search for "GNOME". GNOME 1.0 was released in '99, and was the first Slashdot story with many comments on GNOME, so I looked in there.
Right up near the top, I get "Gnome is for sysadmins - not home users" as a comment title. Searching for "Windows" and "Microsoft" fails to bring up inflammatory comments. I scrolled down, looking at article titles, and saw "not there yet", "Who cares about Joe Sixpack", and "fvwm is the best". I skimmed through the entire first page (I'm not going to do all the pages) and I couldn't see a *single* comment that could be easily identified by its title as being someone claiming that Windows was in trouble or Linux was going to be big on the desktop. Lots of talk about goals, lots of GNOME/KDE bashing (at least some things stay the same), no talking about "this being the year". A lot of people saying that the release was premature.
I think that you're wrong. Most of the talk about Linux was based on the groundswell of support for Linux on servers. There was a lot of "Try Linux! You can use it instead of Windows now!" starting, IIRC, sometime around 2001, but that was on Slashdot and aimed at techies".
KDE libs might have been out in 98, but there was no set of applications to replace Windows apps out.
When people say "this year Linux is going to be big on the desktop", they may be optimistic, but they aren't way out in left field, doing a five-year-early-prediction.
I can't understand what your proposal is, unless you have a very vague concept of where you're going. How are you suggesting one interface with the compiler, describe your requirements? My issue is that it seems that no matter what you do, you're going to be bringing some set of constraints (potentially incorrect) into the ballgame.
As for provability -- CMU has research on proof-carrying Java code, where there are certain things proved about the code. I can also tell you that it is not possible to prove arbitrary things about any piece of code, or you'd have solved the halting problem -- "Will this program get stuck or not?"
Except Miguel never said that C is dead. This was a misquote by the Slashdot article that people are happily throwing around.
:-)
If C is dead, a lot of Miguel's code is going to suddenly disappear.
I can give examples of how they *might* do it. I didn't RTFA either.
* Pointer arithmetic
Pointer arithmetic becomes an operation that calls a function that knows what chunks of memory were allocated and checks before mucking with it.
* Hardcoded type sizes instead of using sizeof() (i.e. assuming sizeof(int) == 4).
I don't see why there would be any problem doing this. This isn't correct C code, though, and hasn't ever been (that being said, C does not natively provide certain guarantees that people require for basic functionality, so I suspect most C code is somewhat broken WRT integer sizes).
* Lax rules for casting
Why would there be a problem? If pointer dereferences are checked, what's wrong with misinterpreting a float as an integer or visa versa?
You know, while I *vastly* prefer to use C-based applications over Java-based ones, I have to say that your example is one of the ones where I would consider Java before C.
You have an application that is presumably a custom job. Developer time is expensive and resource usage is cheap on custom or vertical market jobs. Furthermore, Java does a better job of providing built-in safety checks, and your application deals with money.
If you want programs to dump stack traces, it's pretty easy. You can modify your shell set up to dump out stack traces when a program dies.
Windows 98 has protected memory. You can't do very effective virtual memory in the modern sense of the word without implementing most of what you need to have protected memory.
The main problem that you have, I think, is that a sandbox is a high-overhead system that does fancier error output, handling dynamic typing and the like, killing a program the moment it steps a byte out of line. Since hardware solutions like the MMU aren't *quite* this smart, you can't kill things exactly when they start to behave naughtily.
It would be best if we could move on from procedural/imperative/object-oriented/functional languages to a language based on logic that can be used to prove program goals.
I've thought about this, and pretty much decided that there isn't a really feasible way of doing this. Setting a bunch of assertions and letting the compiler "do the right thing" is, I think, an attractive but ephemeral dream. Why?
Think about what you'd do when, say, sorting a linked list. In a traditional language, you'd shove some pointers around. In a constraint-based language, you'd say something like "after this point, I want this assertion to hold -- for all nodes N1, N2 such that N2=N1->next->next...->next, N2->data <= N1-data". The compiler figures out some sort of efficient way to do up the code. The problem is, you've already imposed a data structure on the compiler, which might not be the right one. What if you want to remove an element from the list? You can do "after this point, I want this assertion to hold -- for all nodes N1 such that N1=head->next->next...>next, N1->data != 3; Then you realize that you just removed an employee from the list before assigning him his final paycheck -- you made a higher-level logic error.
Short of producing a full-blown humanlike AI that guesses and lets you refine, where you can say "Give me a word processor", "Okay, add a spellchecker to it", "Okay, make the menu item to invoke the spellchecker be under the Edit menu from under the File menu, where you dumped it", I don't think it's really feasible to go for really high-level languages -- you just don't get the benefits you think you do.
I've taken a look at aspect-oriented programming, and I've come to the conclusion that it will not catch on all over unless the current programming model radically changes.
The problem with aspect-oriented programming is that it can be very difficult to determine where code is located in your source tree that is causing a particular effect that you're observing at runtime. With a procedural language (and barring extensive use of, say, function pointers), it's generally a fairly straightforward task to locate what code is being run that's causing something to break when you perform operation X.
This can admittedly be solved -- as long as one uses an intelligent IDE that can tell you what code affects what code at runtime when you're poking through source.
The problem is that doing this requires everyone adopting a particular IDE (maybe an extended Eclipse) for effective use.
The programming model that dates back to the earliest compilable programming systems is that you produce text in an ordinary text-editing program, and then compile it. You don't need a special program to work with your project.
It doesn't mean that such a move can't be made, but I do think that AOP will not catch on in general until a move like everyone starting to use Eclipse for software development happens.
If you are writing applications its better to use Java or C# anyday.
You know, I've heard that, and I don't agree, at least as a general rule.
For custom code development or vertical market stuff, it's reasonable, unless there are other significant concerns.
However, an end-product written in C is faster and more memory-efficient. If someone proposed rewriting bash in Java, I'd never use what they put out. Same for the GIMP, xterm, xmms, or any of the other apps that dot my desktop.
Of the Java apps I've tried...I think I've tried straw (RSS reader, found it to be slow and ended up not using it), tube (Java Hotline client, ended up using the C-based fidelio in preference), freenet (still no good C alternative, so I don't run freenet, as it soaks up masses of RAM), and a bunch of other minor apps that just got quickly dropped. If you're writing something that many people are going to be using all over the world, and you want a really polished final product, I claim that you're generally better off with C (and I wish that I could think of a good safe language, but the only safe languages that I've been reasonably impressed with for writing full-blown applications -- eiffel and ocaml -- have a stunning lack of interest from the public).
* C is mature and stable. Yup, there are some bad design decisions, but using C isn't quite like riding the current version of Java -- new idea, whoops, let's change that, oh wait, let's extend that...
* There are good development tools for C. There is a good set of tools that have been built up over the years for analyzing and debugging C.
* C doesn't impose a lot of resource overhead. I'm all for having safe languages, but Java is just plain fat when it comes to memory usage, and encourages a lot of practices (like rapid destruction and creation of objects) that tends to cause a serious CPU hit as well. Safe and *efficient* languages have, as a general rule, flopped in the marketplace.
* C runs on everything. There are awfully few platforms for which there isn't a C compiler.
There are a lot of Java apps out there. Yet, in my day-to-day use of my computer, I don't use a single Java-based program. Not one. I use a small number of perl and python scripts, a couple of bash scripts, and a vast quantity of C code. The C alternatives are faster and more memory-efficient. Plus, I've tended to come to believe that the average Java programmer experience level is lower than than of the average C programmer (simply because lots of Java people *started* coding with Java which puts an upper bound on experience).
I admit that Java can be neat for some things (rapid application development/prototyping, lightweight distributed and network-using apps). It's generally not the sort of thing that sits on my desktop, though.
It may be that more Java code is written each day now than C code, but since so much Java development is vertical market or custom development, I suspect that the overwhelming majority of lines of code *used* by people is C.
Okay, I'm not a DB programmer, so I just may be out of touch here, but I know that Postgres had transactions, I'm quite sure that Oracle had transactions, and I strongly suspect that MS SQL had transaction support already.
If MySQL really did just get transaction support, it seems more like My was a bit behind, rather than being far ahead of the others.
Without HP, Keith Packard wouldn't be getting his paycheck and all of us wouldn't be looking at antialiased fonts.
Well, perhaps nobody else does, but I buy my video cards based on open source support (not Linux support alone).
For years, Matrox had the best support for Linux with open source drivers. I bought Matrox cards. Currently, ATI has the best open source support. Right now, I'm buying ATI. I'll keep doing so, as well. I use my cards under Linux exclusively, and binary drivers are a tremendous pain in the ass to deal with. I recognize that video card vendors have reasons for wanting to keep their drivers closed-source -- that's fine, but I happen to value open source.
Depends on how you chose to set it up (or in the case of Joe Six-pack, how the packager does). It's not that hard to grant ordinary users write privs to the CD device.
You're full of it. You are trying to say, with a straight face, that people expected Linux to take over the desktop in *'98*?
Most of the time in the past was people getting excited about the ability for *geeks* to use exclusively Linux -- Open Office, Samba, etc made it feasible to work with Windows users and still keep using Linux.
Red Hat's CEO said, what, six months ago that Linux isn't ready for the desktop war just yet?
This year and last year are big because there are a lot of major open source apps coming out and being *usable*, by *typical users*, at at least a basic level, as a substitute for Windows apps.
Finally, if you don't think Linux usability has improved massively since '98, you just plain don't remember 98. We had no GNOME or KDE apps. Preference dialogs didn't exist. Widget sets were Tk, and black-and-white Athena. Boxes required a serious sysadmin to secure out-of-box.
Last year, I agree that there were a lot of people on Slashdot that were predicting big gains on the desktop. And guess what? A bunch of governments and big companies starting transition processes, or at least made it much more easy to move a chunk at a time to Linux. If anything, I'm surprised that things are going this quickly.
My prediction is that Linux will break 10% desktop market share before the end of 2006. That is a *huge* number of users to move from one platform to another -- perhaps around 100 million users -- , but remember that there's a threshhold effect at which point application vendors, people doing file formats, etc cannot ignore Linux, and once that hump is over, it becomes much easier to move to Linux.
Web sites are already improving -- I don't see the number of "IE-only" sites that I did thanks to the spread of Mozilla, Linux, and Mac OS X running Safari.
That being said, I think that as Microsoft gets more worried, they will do whatever it takes to fight back effectively. That may be as far as moving to a Linux-based distribution and porting their products to it. Microsoft is unlikely to die, no matter what.
I don't believe that 20% of all corporate desktops in the world will be running Linux by the end of 2005. No, that just isn't plausible.
Businesses do transitions when they bring in new machines. A typical business machine has, what, a four year lifecycle? That means that even if every machine being replaced was converted to Linux, there'd be less than 50% people using Linux. You're requiring about half of all new desktop purchases to be Linux-based. That's wildly unrealistic.
I'd say that 20% Linux desktop business penetration by the end of 2007 would be a very positive outcome.