The only thing I worry about as far as Linux is concerned is the desire that some people express for the one trueanything. Competition fosters evolution through rivalry and through cross pollination. In my opinion, GNOME would not have existed (or at least been developed as quickly) without KDE. Conversely, I doubt that Troll Tech would have worked as hard on the QPL if not for pressure brought on by GNOME.
Standards are fine, and applications should play together, but software should evolve. I run Window Maker with GNOME and KDE applications just fine, thanks.
I want to be able to choose between distributions, desktops and applications. Luckily, as long as the code is free (as in speech), I'll always have a choice. Welcome to the revolution, enjoy the ride.
Nick Petreley has been promoting Linux for years in Info World. He was supporting Linux long before it was fashonable and doing so in a clueful manner. He supports Linux because he thinks it works and is far too well known and respected to need to brown nose anybody.
I don't know where you get your information, but you should find a more reliable source.
I'll take your post as unexaggerated truth. My experience is that the vast majority of people can not set up an NT server that is reliable. So I would guess that you are either amazingly lucky or highly skilled and trained in the art of setting up reliable NT servers. Linux servers are reliable "out of the box." Red Hat servers are not overly secure out of the box (too many services running and anonymous FTP), but NT is neither secure nor stable in its default configuration. So my first point is that a moderately trained Linux administrator can set up a secure and reliable Linux box, whereas my experience suggests that it takes someone of extraordinary skill to set up a stable and secure NT box.
I only dabble with web servers so I can't speak with a great deal of knowledge on the subject. I have seen a plenty of web sites created with PHP, mod_perl and Zope that seem pretty cool, although it may have taken longer for them to be developed, I don't know. But IBM is porting Web Sphere to Linux, a port of Cold Fusion is on the way and other vendors are porting tools to Linux practically everyday now. Therefore, my second point is that as far as RAD development is concerned, I don't imagine that NT will have much of an advantage in the future, assuming that it has an advantage now.
The mere fact that you can run a Linux without a GUI gives a major performance boost as a server. Although I only dabble with web servers, I know operating systems. I started my career supporting a proprietary, multitasking operating system (CTOS) in the early 80s. I've written transport layers on top of DOS, multithreading toolkits for DOS and Windows 3.1 and device drivers for ISC Unix, Linux and (shudder) NT. I can say without any reservation that NT is incredibly inefficient. The number of "objects" that you need to pass around in a device driver is absurd, not to mention all the painful abstractions and IRQ layers that you need to worry about. The funny aspect is that all the inefficiencies do nothing to improve the stability of the OS. As a matter of fact, a bug in a device driver is guaranteed to BSOD the system. In Linux, there is at least a chance that a bad sound driver won't crash the entire system. Point number three, a Linux server will almost always be faster and more stable than an NT server.
Now although you set up NT boxes for your clients that "never crash", what do you do something does go wrong? I'm assuming it's the client's fault, maybe they installed the latest Internet Explorer service pack and now their server is unstable (this really happened to me). For an NT machine, you drive to the client site and start mucking in the registry, checking DLL versions and pointing and clicking the night away. If they don't have a good disaster recover mechanism in place (including backing up the registry), you back up your software, format the hard drive, and reinstall everything from the ground up. What do you do with a Linux box? Well first of all, the latest web browser won't make a Linux server unstable (it may crash X, but it won't crash the system). But things can go wrong and when they do, you secure shell in, and go through a similar (but in my opinion, less painful) process. You can even reboot using a different kernel if you need to. All the while you're in your pajamas listening to your favorite CDs. Only if a Linux server is completely hosed do you need to be physically present. So the fourth point is that Linux is easier to support.
Finally, scalability. If the customer's machine is under powered and needs to be upgraded, Linux scales better. On Alpha and UltraSparc hardware, Linux is 64-bit. IBM will be supporting Linux on RS/6000s in the very near future. Besides that, any solution you develop for Linux will easily port to any UNIX out there in case they really need a Sun Enterprise server with a ridiculous number of processors or an HP 9000. The customer with the NT solution is screwed until Merced and NT 6.0 (2005?) ship. Trust me, they won't be happy. Point five and game goes to Linux.
Can you say Freedom? I hope you can.
P.S. I just reread this and it seems a bit pompous. I didn't mean it to be, it must be that I watched "Pride and Prejudice" last night. Oh well, I'm too lazy to rewrite it.
You guys need to get your name out. I've been hearing about LinuxCare for months now, whereas I didn't know about you guys until recent posts to Slashdot. Or have I been blind, deaf, and dumb. The dumb part is especially likely, although not really relevant.
PGCC is a commercial C++ compiler that supports precompiled headers. They also have an NT version as well as a Linux (and other UN*Xes) version which could make your cross platform issues easier to deal with.
Remember, gcc != Linux. Please consider not insulting your target audience next time you ask for help.
So if everyone thought that SuSE was "GNU/Linux" you'd be happy? Doesn't SuSE ship with KDE 1.x which is based on Qt 1.x (i.e. the version of Qt with the objectionable license)? How unGNU...
SAGLE (SuSE A'int GNU/Linux Either)
If you need a "holy-grail" of GNU and Linux, Debian and Slackware are the only choices... But Red Hat is way more in line with FSF philosophies than any other for-profit distribution that I'm aware of.
In case you care, I run Red Hat although I'll probably give Debian 2.1 a spin.
The word "binary" never appears in the article and it clearly states that HP intends to use the Open Source development model. It said that PA-RISC development was being done by the Puffin group, so I'm really confused by your "only binaries" comment.
Of course Red Hat is going to be the first choice for a US company. Red Hat has significant name recognition and they've been scrambling to get 24x7 support in place (that's what they are doing with the money from Intel and Netscape). The article also pointed out that HP was going to use Linux distributors for there support. About the only other choice today in the US for 24x7 is Caldera. Finally, HP said that their agreement with Red Hat is not exclusive. Until LinuxCare and similar multi-distribution support centers get ramped up, companies are going to go with the big distributions in their region (Red Hat in the US, SuSE in Europe and Pacific High Tech in Asia). Hopefully will get large companies to recognize Debian in the future, right now there isn't an accepted 24x7 solution for it.
This post could get ridiculously long, since this is a topic near and dear to my heart. I'll try to keep it short...
I think it is a mistake to compare Open Source with communism or even socialism. At least in implementation, communism (and socialism to a lesser degree) tends to have a central bureaucracy that decides how all resources will be distributed. This is extremely inefficient and prone to error. You end up with an abundance of things that people don't want and shortages of things that people want. Why, because it takes time to convince a bureaucracy that they "got it wrong." What ever "it" is. Especially because you have some central authority that knows what's best for everyone, even if everyone happens to disagree.
Another problem with communism is that the central authority is forcing its subjects to redistribute their wealth (goods, service, skills, resources, etc.) based on the central authority's desires, not the producer's desires. This is the point you made, it's against human nature to work hard for someone else's benefit.
Open Source is a completely different beast. The argument about human nature doesn't come directly into play. The reason is that contributers are working on something they care about, directly. Open Source is not an abstract goal (scrubbing toilets for the good of humanity). Open Source is created largely for the benefit of the person contributing. Why does Apache exist? Because its authors wanted a good Web Server, period. Why share it? There are probably too many reasons to list, but I'll give the usual short list: Pride, recognition, and as a gift. After all, greed is only one facet of human nature. But, the key point is that the contributer selects work that is beneficial for their own personal reasons.
As far as efficiency, Open Source is efficient in a mind bogglingly chaotic manner. The products are generated in direct response to a need. It really reminds me of evolution. Several products may be generated for the same or similar needs. The products compete for all sorts of resources; contributers, users, distribution channels and good old fashion "mind share." Products that can get adequate resources flourish and evolve, products that can't stagnate.
It's late, hopefully my ramblings make some semblance of sense.
From the article: "The kernel" -- Linux's most vital code -- "is 1 percent of the entire program," he says. "Of that 1 percent, I've written between 5 and 10 percent. I think the most important part is that I got it started. Then people had something to concentrate on." Indeed, Torvalds places the number of volunteers who regularly contribute to the "kernel" at about 1,000, and thousands more have sent in pieces of code over the years.
And before you jump all over the Linux's most vital code comment, that was clearly made by the author. My pet peeve is when people use a single, out of context statement to air their pet peeve.
I've been assuming, quite jadedly, the MS wouldn't be able to produce the code and environment to rebuild identical binaries (even if they wanted to). Presumably Caldera has already presented the disassembled section of code in question, in less of course they felt bound by the EULA.:-)
Whether or not DR-DOS is a viable product today is not the issue. The issue is whether MS used illegal market barriers to prevent DR-DOS from competing with MS-DOS. If Caldera can prove that MS prevented DR-DOS from gaining market share illegally, then MS owes Caldera damages. Although Caldera was not the company that developed DR-DOS, each time DR-DOS has changed hands, part of the package was any potential damages won from MS.
I believe that there is actually a strong case against MS. Before MS's per-CPU licensing schemes DR-DOS was gaining market share. The combination of per-CPU market share and Win 3.1 refusing to run on DR-DOS, killed DR-DOS. The 1995 ruling against MS declared per-CPU licensing illegal, so that half of their argument seems solid. I don't know if they'll get anywhere with the source code arguments, I'd think MS would remove any damaging code before turning over the source.
The whole point of Debian is to produce a 100% free distribution. It would be a bit silly to allow people to join who intended to subvert the fundimental purpose of the distribution.
Having the maintainers agree to contribute code using licenses that meet thier criteria for "free" software in has nothing to do with technical issues. Debian is not saying, "all programs must be single threaded" or "GUIs are for wimps, end of discussion." They are saying that they will only build a their distribution using licenses that meet specified requirements. That doesn't stiffle the discussions and arguments about technical issues, usability concerns or any other tangible aspect of the distribution. It just guarantees the consumers of the distribution that they can do anything they want with the distribution, except make the code secret.
I assume you are implying that contracting Alan Cox to hack the kernel is not actually useful... Yes, I know what you're real point was and all I can say is that competition is good.
Neither KDE nor GNOME exclude the use of each others applications. They've adopted the same drag and drop protocol. And I believe that most viable window managers will eventually support hints from both kind toolkits (as WM already does). Furthermore, Gtk+ is themeable and KDE will be when Qt 2.0 is released. By the time KDE and GNOME have there next major revisions I'd bet you'll be able to select a theme (or two) that will make it hard to tell which apps are using which toolkit. You'll be able to tell, but the novice user wouldn't even notice.
As for consistency across the desktop, MS isn't consistent with its use of menus, shortcuts and mouse clicks across the Office applications, much less the set of applications that ship with Windows 9x/NT and from third party vendors. I know, your example was BeOS and Mac, but winning the hearts of MS users wouldn't be to shabby of a start.
Personally I'm glad there are several Open Source desktop/Window Manager projects (everyone seems to forget about GNUStep and tkDesk), diversity accelerates evolution.
BTW, DNRH (Debian's Not Red Hat - and they support GNOME).
There was already a poll, Katz won. It is not a majority of/.'ers that don't like Katz. Flaimers are just noisier than lurkers, that does not make them the majority.
You're right that most people are perfectly happy with Microsoft products. But what you don't seem to be aware of is how MS structures its agreements with OEMs to prevent competing products from ever having a chance. The margins in commodity PCs are extremely small, and if you sell PCs most of your customers are going to want Windows. Back when DR-DOS was gaining popularity, MS struck deals with OEMs that gave them major discounts if they paid the discounted MS-DOS license for every CPU shipped, even if MS-DOS wasn't installed. Since price was the primary point of competition and since most (not all) customers wanted MS-DOS, the OEMs really did not have a choice. They had to sign the per-CPU agreements or go out of business. They'd still sell you a machine with DR-DOS, but it would cost $50 more than the machine with MS-DOS because you were paying for both OSes. Before the per-CPU licensing went into effect, DR-DOS was gaining significant market share. With the per-CPU agreements and some other maneuvering, MS killed DR-DOS. The 1995 ruling against MS made it's per-CPU agreements illegal.
So after 1995, it really was the free market that kept out would be competitors, right? Wrong. MS restructured its agreements with OEMs such that rather then charging for Windows 9x per-CPU, they charge a licensing fee per "model." If Toshiba were to install BeOS on a 650CT, they'd still have to pay a license fee for Windows 9x. To avoid paying the license fee they'd have to introduce a new model, which means silk screening and manuals, which just isn't worth it for a niche market. But it gets better. Besides tying the license fee to a model number, the discount is also dependent on a high percentage (I don't know the number) of all systems they sell shipping with an MS OS. There is no incentive on the OEMs part to expand their offerings in alternative OSes, because they would lose their licensing discount in their primary market if the alternatives were even marginally successful. The large OEMs are finally starting to offer Linux on their servers. The only reason they can do this is because of relatively low volume of servers, higher margins, and the high cost of Windows NT server. I don't think you'll be seeing BeOS, Linux or FreeBSD on Dell's low-end machines, at least not without MS getting it's cut. MS has its hands tied with the antitrust case, but if they win the case, you can expect that MS will start turning the economic thumb screws again.
I'm amused by your comment about MS's practice of forcing products on OEMs so they can get Windows as "for the most part stopped." In what way has it stopped? MS told Gateway that it could not ship Netscape as the primary browser (with its icon on the desktop instead of IE's icon). I have no problem with the browser being part of the desktop, but there is no technical reason that prevents IE from being part of the desktop and Netscape being you're default web browser on the desktop. Are you going to tell me that Gateway had a choice? Sure, they could have chosen to go out of business.
Have you read you're EULA for any MS applications? The EULA states that you are only licensed to run the application on an approved MS OS. It doesn't say they won't support the application on non-MS OSes (which would make sense), it says you are not allowed to run the application on other platforms. There aren't any non-MS OSes that can run MS Office, but MS has made it illegal just in case. The list of abuses of the market and thier customers goes on and on.
MS has no interest in the free market, they throw up roadblocks at every turn. Those of us who don't want any of our money going to MS's war chest have every right to be pissed off and to want our money back.
I almost agree. I think the DOJ should not focus on the browser issue because, as you point out, a browser is becoming a required tool on the desktop. The issue is whether Microsoft is a monopoly, and if so, do they use their monopoly position illegally. I believe it is hard to deny that Microsoft has a monopoly in the Desktop OS market. Given the facts that they: 1. Replaced their per CPU licenses with per model licenses. 2. Intentionally polluted Java in an attempt to lock developers to the Win32 environment. 3. Forced OEMs to bundle IE with Windows 95, which clearly did not require IE. 4. Prior to #3, MS explicitly told Netscape that if they didn't play ball, MS would crush Netscape. The list goes on, but I it seems clear to me that MS illegally exploits their monopoly position.
I don't think the major issue in the DOJ case is that MS "gave away" the browser. The issue is that MS forced OEMs to use IE as the primary browser on the desktop (especially for Windows 95 OSR2). Microsoft's assertion that IE is part of the OS is absurd, IE is an application that is used by the desktop and other MS applications. It would be fine to say that in order for the MS desktop to work, you need these IE DLLs, but that's not what MS did. They said "if you don't configure windows to use IE as the default browser, you can't ship Windows." Compaq, Dell, etc., should be allowed to ship Netscape as the primary browser if they choose.
I need Gtk+ to run Gnome, that does not make Gimp "part of the OS." I hope MS is found guilty of exploiting their position and I hope the US government stays out of the OS design business...
Slashdot Mirror
Standards are fine, and applications should play together, but software should evolve. I run Window Maker with GNOME and KDE applications just fine, thanks.
I want to be able to choose between distributions, desktops and applications. Luckily, as long as the code is free (as in speech), I'll always have a choice. Welcome to the revolution, enjoy the ride.
I don't know where you get your information, but you should find a more reliable source.
"World Domination" is a joke. It is specifically aimed at MS, tongue in cheek.
I only dabble with web servers so I can't speak with a great deal of knowledge on the subject. I have seen a plenty of web sites created with PHP, mod_perl and Zope that seem pretty cool, although it may have taken longer for them to be developed, I don't know. But IBM is porting Web Sphere to Linux, a port of Cold Fusion is on the way and other vendors are porting tools to Linux practically everyday now. Therefore, my second point is that as far as RAD development is concerned, I don't imagine that NT will have much of an advantage in the future, assuming that it has an advantage now.
The mere fact that you can run a Linux without a GUI gives a major performance boost as a server. Although I only dabble with web servers, I know operating systems. I started my career supporting a proprietary, multitasking operating system (CTOS) in the early 80s. I've written transport layers on top of DOS, multithreading toolkits for DOS and Windows 3.1 and device drivers for ISC Unix, Linux and (shudder) NT. I can say without any reservation that NT is incredibly inefficient. The number of "objects" that you need to pass around in a device driver is absurd, not to mention all the painful abstractions and IRQ layers that you need to worry about. The funny aspect is that all the inefficiencies do nothing to improve the stability of the OS. As a matter of fact, a bug in a device driver is guaranteed to BSOD the system. In Linux, there is at least a chance that a bad sound driver won't crash the entire system. Point number three, a Linux server will almost always be faster and more stable than an NT server.
Now although you set up NT boxes for your clients that "never crash", what do you do something does go wrong? I'm assuming it's the client's fault, maybe they installed the latest Internet Explorer service pack and now their server is unstable (this really happened to me). For an NT machine, you drive to the client site and start mucking in the registry, checking DLL versions and pointing and clicking the night away. If they don't have a good disaster recover mechanism in place (including backing up the registry), you back up your software, format the hard drive, and reinstall everything from the ground up. What do you do with a Linux box? Well first of all, the latest web browser won't make a Linux server unstable (it may crash X, but it won't crash the system). But things can go wrong and when they do, you secure shell in, and go through a similar (but in my opinion, less painful) process. You can even reboot using a different kernel if you need to. All the while you're in your pajamas listening to your favorite CDs. Only if a Linux server is completely hosed do you need to be physically present. So the fourth point is that Linux is easier to support.
Finally, scalability. If the customer's machine is under powered and needs to be upgraded, Linux scales better. On Alpha and UltraSparc hardware, Linux is 64-bit. IBM will be supporting Linux on RS/6000s in the very near future. Besides that, any solution you develop for Linux will easily port to any UNIX out there in case they really need a Sun Enterprise server with a ridiculous number of processors or an HP 9000. The customer with the NT solution is screwed until Merced and NT 6.0 (2005?) ship. Trust me, they won't be happy. Point five and game goes to Linux.
Can you say Freedom? I hope you can.
P.S. I just reread this and it seems a bit pompous. I didn't mean it to be, it must be that I watched "Pride and Prejudice" last night. Oh well, I'm too lazy to rewrite it.
...Nice Try.
You guys need to get your name out. I've been hearing about LinuxCare for months now, whereas I didn't know about you guys until recent posts to Slashdot. Or have I been blind, deaf, and dumb. The dumb part is especially likely, although not really relevant.
Here's a link to information on the ColdFire processor. It's basically a 68K core targeted at the embedded market.
Remember, gcc != Linux. Please consider not insulting your target audience next time you ask for help.
SAGLE (SuSE A'int GNU/Linux Either)
If you need a "holy-grail" of GNU and Linux, Debian and Slackware are the only choices... But Red Hat is way more in line with FSF philosophies than any other for-profit distribution that I'm aware of.
In case you care, I run Red Hat although I'll probably give Debian 2.1 a spin.
Try ELKS. :-p
Of course Red Hat is going to be the first choice for a US company. Red Hat has significant name recognition and they've been scrambling to get 24x7 support in place (that's what they are doing with the money from Intel and Netscape). The article also pointed out that HP was going to use Linux distributors for there support. About the only other choice today in the US for 24x7 is Caldera. Finally, HP said that their agreement with Red Hat is not exclusive. Until LinuxCare and similar multi-distribution support centers get ramped up, companies are going to go with the big distributions in their region (Red Hat in the US, SuSE in Europe and Pacific High Tech in Asia). Hopefully will get large companies to recognize Debian in the future, right now there isn't an accepted 24x7 solution for it.
Sheesh, mention Red Hat and people freak out...
I think it is a mistake to compare Open Source with communism or even socialism. At least in implementation, communism (and socialism to a lesser degree) tends to have a central bureaucracy that decides how all resources will be distributed. This is extremely inefficient and prone to error. You end up with an abundance of things that people don't want and shortages of things that people want. Why, because it takes time to convince a bureaucracy that they "got it wrong." What ever "it" is. Especially because you have some central authority that knows what's best for everyone, even if everyone happens to disagree.
Another problem with communism is that the central authority is forcing its subjects to redistribute their wealth (goods, service, skills, resources, etc.) based on the central authority's desires, not the producer's desires. This is the point you made, it's against human nature to work hard for someone else's benefit.
Open Source is a completely different beast. The argument about human nature doesn't come directly into play. The reason is that contributers are working on something they care about, directly. Open Source is not an abstract goal (scrubbing toilets for the good of humanity). Open Source is created largely for the benefit of the person contributing. Why does Apache exist? Because its authors wanted a good Web Server, period. Why share it? There are probably too many reasons to list, but I'll give the usual short list: Pride, recognition, and as a gift. After all, greed is only one facet of human nature. But, the key point is that the contributer selects work that is beneficial for their own personal reasons.
As far as efficiency, Open Source is efficient in a mind bogglingly chaotic manner. The products are generated in direct response to a need. It really reminds me of evolution. Several products may be generated for the same or similar needs. The products compete for all sorts of resources; contributers, users, distribution channels and good old fashion "mind share." Products that can get adequate resources flourish and evolve, products that can't stagnate.
It's late, hopefully my ramblings make some semblance of sense.
And before you jump all over the Linux's most vital code comment, that was clearly made by the author. My pet peeve is when people use a single, out of context statement to air their pet peeve.
I've been assuming, quite jadedly, the MS wouldn't be able to produce the code and environment to rebuild identical binaries (even if they wanted to). Presumably Caldera has already presented the disassembled section of code in question, in less of course they felt bound by the EULA. :-)
I believe that there is actually a strong case against MS. Before MS's per-CPU licensing schemes DR-DOS was gaining market share. The combination of per-CPU market share and Win 3.1 refusing to run on DR-DOS, killed DR-DOS. The 1995 ruling against MS declared per-CPU licensing illegal, so that half of their argument seems solid. I don't know if they'll get anywhere with the source code arguments, I'd think MS would remove any damaging code before turning over the source.
I wouldn't mind a $1.6 billion dead horse. :-)
Having the maintainers agree to contribute code using licenses that meet thier criteria for "free" software in has nothing to do with technical issues. Debian is not saying, "all programs must be single threaded" or "GUIs are for wimps, end of discussion." They are saying that they will only build a their distribution using licenses that meet specified requirements. That doesn't stiffle the discussions and arguments about technical issues, usability concerns or any other tangible aspect of the distribution. It just guarantees the consumers of the distribution that they can do anything they want with the distribution, except make the code secret.
In closing, go SGI!
As for consistency across the desktop, MS isn't consistent with its use of menus, shortcuts and mouse clicks across the Office applications, much less the set of applications that ship with Windows 9x/NT and from third party vendors. I know, your example was BeOS and Mac, but winning the hearts of MS users wouldn't be to shabby of a start.
Personally I'm glad there are several Open Source desktop/Window Manager projects (everyone seems to forget about GNUStep and tkDesk), diversity accelerates evolution.
BTW, DNRH (Debian's Not Red Hat - and they support GNOME).
There was already a poll, Katz won. It is not a majority of /.'ers that don't like Katz. Flaimers are just noisier than lurkers, that does not make them the majority.
So after 1995, it really was the free market that kept out would be competitors, right? Wrong. MS restructured its agreements with OEMs such that rather then charging for Windows 9x per-CPU, they charge a licensing fee per "model." If Toshiba were to install BeOS on a 650CT, they'd still have to pay a license fee for Windows 9x. To avoid paying the license fee they'd have to introduce a new model, which means silk screening and manuals, which just isn't worth it for a niche market. But it gets better. Besides tying the license fee to a model number, the discount is also dependent on a high percentage (I don't know the number) of all systems they sell shipping with an MS OS. There is no incentive on the OEMs part to expand their offerings in alternative OSes, because they would lose their licensing discount in their primary market if the alternatives were even marginally successful. The large OEMs are finally starting to offer Linux on their servers. The only reason they can do this is because of relatively low volume of servers, higher margins, and the high cost of Windows NT server. I don't think you'll be seeing BeOS, Linux or FreeBSD on Dell's low-end machines, at least not without MS getting it's cut. MS has its hands tied with the antitrust case, but if they win the case, you can expect that MS will start turning the economic thumb screws again.
I'm amused by your comment about MS's practice of forcing products on OEMs so they can get Windows as "for the most part stopped." In what way has it stopped? MS told Gateway that it could not ship Netscape as the primary browser (with its icon on the desktop instead of IE's icon). I have no problem with the browser being part of the desktop, but there is no technical reason that prevents IE from being part of the desktop and Netscape being you're default web browser on the desktop. Are you going to tell me that Gateway had a choice? Sure, they could have chosen to go out of business.
Have you read you're EULA for any MS applications? The EULA states that you are only licensed to run the application on an approved MS OS. It doesn't say they won't support the application on non-MS OSes (which would make sense), it says you are not allowed to run the application on other platforms. There aren't any non-MS OSes that can run MS Office, but MS has made it illegal just in case. The list of abuses of the market and thier customers goes on and on.
MS has no interest in the free market, they throw up roadblocks at every turn. Those of us who don't want any of our money going to MS's war chest have every right to be pissed off and to want our money back.
1. Replaced their per CPU licenses with per model licenses.
2. Intentionally polluted Java in an attempt to lock developers to the Win32 environment.
3. Forced OEMs to bundle IE with Windows 95, which clearly did not require IE.
4. Prior to #3, MS explicitly told Netscape that if they didn't play ball, MS would crush Netscape.
The list goes on, but I it seems clear to me that MS illegally exploits their monopoly position.
I don't think the major issue in the DOJ case is that MS "gave away" the browser. The issue is that MS forced OEMs to use IE as the primary browser on the desktop (especially for Windows 95 OSR2). Microsoft's assertion that IE is part of the OS is absurd, IE is an application that is used by the desktop and other MS applications. It would be fine to say that in order for the MS desktop to work, you need these IE DLLs, but that's not what MS did. They said "if you don't configure windows to use IE as the default browser, you can't ship Windows." Compaq, Dell, etc., should be allowed to ship Netscape as the primary browser if they choose.
I need Gtk+ to run Gnome, that does not make Gimp "part of the OS." I hope MS is found guilty of exploiting their position and I hope the US government stays out of the OS design business...