What you say true that if you assumme you can make a profit from selling GPL code (by either selling support, or copyrighted addon's, or just encouraging customers to buy the original version rather than a copy). TrollTech does assumme this, and yes they had to modify the GPL to their license that disallows redistribution for any profit, even of a GPL version.
However the vast majority of people, especially everybody who does not like the GPL, claim that it is a fact that you cannot make a profit selling GPL code (even though I kind of like the GPL, I can agree with this somewhat, people are cheapskates and if the free copy is legal they will get it). Therefore if you make this assumption you can assume that only the original author can profit from the GPL code, since only they can dual-license it into a form that you can make a profit from.
It is refreshing to see that even an anti-GPL advocate can admit you can make money using the GPL, however. You seem more convinced of it than I am.
You are contradicting yourself. You say that "using the GPL is harmful to developers" which implies that developers have an interest in aquiring a non-GPL version of a program. You then say that the fact that TrollTech releases a GPL version harms their own business by removing the incentive for people to get the commercial version.
A more practical response is that all theoretical damage of the GPL is well known and is avoided in the real world. All code of any value for developing new and innovate programs is released using licenses like the LGPL, LGPL with exceptions to make it even more like public-domain, or BSD or public-domain licenses.
I write LGPL+exceptions code (the exceptions are the typical ones where you can use the library as though it was a public-domain closed source library, but if you modify any of the source code to the library you must release all the changes if you redistribute it). I know for a fact that my code would not be used if it was GPL so I am strongly encouraged to not do that. This is true of almost any writer of open-source code that is designed to be a service, and thus completely defuses any argument that it somehow harms advancement and use by anybody. I also write GPL code but that is because it is an end application, that code is useless to anybody except to make a very similar program, which I would consider the theft of my ideas and IP. Also I know that the fact that it is GPL has zero impact on the popularity of this code, because it has zero impact on the ability to run it, which is the main use.
I agree with most of what you say very much, but your claim that the GPL does not help the author is false. You apparently have not heard of dual licensing, which is something the original author can do. The GPL version of the code can be an excellent and very inexpensive advertisement for the code. But the author can still sell the rights to use the code in closed-source products, or may just keep some enhancements to the code secret and sell those.
Check how TrollTech is doing and their business model for details.
The reason that that GPL does not force you to release changes to the code if you don't distribute it is practical:
The GPL is designed entirely to be an "exception" to normal copyrights. If you ignore or "disobey" the GPL then all that is left is normal copyright laws, which lets you do *less* than the GPL allows. Therefore there is zero reason to "disobey" the GPL because you gain nothing and you lose a slight amount (perhaps nothing if you did not plan to redistribute anyway).
Because the GPL does not restrict you from doing anything you normally could do with copyrighted code, it cannot be considered a "contract". Thus it cannot be challenged with any "I didn't agree to that" arguments that make EULA and click-to-agree type contracts legally useless.
Now normally when you buy a book, you are allowed to scribble all over it for any reason you want. Copyrights and other laws do not prevent that. So the GPL has to allow what copyright does, so in no way can it be considered a restriction.
Now to be fair, the GPL could say that you must give RMS your first-born. If you chose to disobey that part of the GPL, and a court agreed with you, you would still be bound by copyright and unable to copy the GPL code, but you would not have to give up your first-born. So adding restrictions would in theory not harm the GPL because legally at worst it would end up where it is now. However in the court of public opinion, having *any* part of it challenged would cause people to think the whole thing is meaningless.
This is also why "advertising" clauses are not allowed by the GPL, and why there are no restrictions allowed on what the software is used for.
The point he was making is that if Sun was allowed to "enhance" the code with the AT&T code, and they were allowed to distribute the result, then they must have distributed the result under the GPL if they did not violate the IP rights of all the other contributors to the Linux kernel. This would mean this valuable code is now available under the GPL and can be used in any other copies of Linux, including ones from IBM and even from SCO.
I was going to say this is unlikely, but in fact it is quite possible that Sun did everything legally, and for the benifit of all parties. AT&T got money for the rights and may still retain the rights for use of the code in non-GPL products. Sun paid money for the rights to use the code in their products, and by adding to the Linux code insured compatability and extensive testing of this code, and avoided the expense of supporting two different code sets. Most likely both AT&T and Sun got some other code that is valuable that did not go into Linux. And of course Linux got some useful code.
Actually I think it is totally obvious that SCO is acting in a way that has nothing to do with them winning money but instead in publicising a "Linux is dangerous" piece of FUD as widely as possible. You would think that if they plan to collect money from Linux they would be encouraging as much use as possible, for instance by saying nonprofit and educational uses are free, and they only want to collect licenses from people selling it or using it for profit. If they succeed in making Linux impossible to use their income from licenses would be zero, obviously, so it makes no sense that they would do all this negative rhetoric. Also claiming that the stuff they say is their own IP is "crap" is really stupid if you expect to collect money for it.
This leads to the obvious conclusion that Microsoft is funding them. And they have made a big point that Microsoft bought a license from them at an apparently very inflated price (many times what Sun paid apparently, does anybody know how much Sun paid?)
However there is another possibility: SCO is making people believe that Microsoft is paying them. They want people to believe they have this new business plan where they will collect money from Microsoft as payment for making press releases, and this is a stable source of huge profits. In some ways it makes a lot of sense: if Microsoft is paying them, money invested in SCO stock a month ago and sold now would collect you a whole lot more of the profits of Microsoft than if you had invested in Microsoft stock directly.
I seriously belive that most, if not all, the outside investors who are buying their stock belive this, and are hoping for Microsoft to purchase more licenses from SCO.
Just want to mention again: showing the code and having it removed from the kernel will help their case, not hurt it. They can then point at the fact that it was removed as proof that the Linux hackers belived the code was stolen, and their lawsuit against the actual copyright infringer would be easy.
The fact that they will not reveal any of the code proves that they have no case and their purpose is to generate FUD about Linux.
I don't think all the investors are stupid. What nobody has mentioned is that not only Linux fanatics believe that Microsoft is funding SCO. Investors may be predicting that there will be a continuing series of "sales" of "product" to Microsoft and that SCO will actually make income that way.
Some of SCO's actions seem even designed to help investors believe this. Most of the press releases clearly claim "Linux is crap" which is pretty foolish if you actually want people to pay you for it rather than go with something else.
Absolutely. The problem is that they don't believe it.
But if they provided cygwin-like functionality by default in Windows they would eliminate 50% of the reasons people work with Linux. If they made it easy to port skills back and forth amoung the systems they would get rid of a huge roadblock to integrating Windows into Unix environments and to assimilating people with Unix skills. And they can compete in a *nice* way: add a "feature" that is useful, but not vital, so portable programs can ifdef it out, and just be less fancy. Document that feature, or even better provide a toy version that runs on Linux, so users can be pretty certain that the interface for that feature is defined and may be portable in the future. Doing this will kill any incentive for Linux programmers to make a different implementation, so the Windows one will be obviously better. This would be successful competition, but will also completely defuse the hate most programmers (like me) have towards Microsofts actions.
Linux advocates don't like to admit it, but if NT had appeared in 1990 with cygwin (or even Xenix) functionality, Linux would be nowhere today and Microsoft would have the total monopoly that they wanted. Oddly enough though, Microsoft refuses to admit this either, somewhere deep in Bill Gates and the NT designer's psyches they know they blew it by ignoring standards, but they are loathe to admit it, even though that would be the smartest thing they could do for their company.
Try fltk for a demo of how this works. The "menu" demo program has a popup menu on the right mouse button that pops up with the cursor pointing at the previously selected item.
Vertical scrolling is currently done by moving the entire menu window, not it's contents. This happens when the mouse is moved to the edge of the screen where the menu goes off the screen. This does have some problems, for instance to stop the scrolling the user must move the mouse at least 1 pixel away from the edge. Horizontal scrolling is not implemented yet. I think the scrolling behavior I have needs work and testing, also X is not really all that hot at moving windows, for instance I only move one window, not the whole hierarchy, because it won't move them nicely.
I have not figured out any way to handle the popup menu whereever the mouse is without placing some of them off-screen. Moving the mouse pointer is bad UI, I tried it, and tried moving it back after release. This was not very user-friendly.
You idiot, if you read my post you would see that I was complaining that perhaps not the naming, but the structure needs to be changed. This is huge compared to adding a symbolic link that Linux defenders keep suggesting will "fix" Linux. I am actually saying that much bigger changes are needed than morons who think long english words are a way to fix bad structure.
You are the idiot for thinking that anybody who suggests anything different than the Microsoft line is not being imaginative.
And I will stand by my suggestion that short abstract names are better. They do not convey meaning that can confuse people about the real purpose, and they are much easier to type and to communicate vocally. Does "Program Names" have a space in it or not? Why not call it "Programs". And why are there things in there that are not programs? And what do French people think about that name? In my experience *nobody* misspells "bin" and it can be clearly stated over the phone and in email.
The problem with pop-up at mouse windows is that all three buttons have uses in most apps now (click, paste, concise (context?) menu). So you eaither need to add a keybord stroke or fundimentily redesign and lose good functionality.
This however is the reason behind most of the inconsistency of Linux (and Windows!) applications. Why doesn't ctrl+s save in all programs? Mostly because a lot of programs assigned ctrl+s to do something else and people don't want to change the interface.
I personally believe the mouse should have a dedicated menu key, with the word "Menu" printed on it (or a little picture, but then the menus cannot change to pies or other shapes). And programs should not use it for anything other than menus.
Position of menus when near edge of screen
My solution to this is to pop it up in the correct position, with it going off the screen, and then have autoscroll that moves it in if you move the mouse to the edge. This also fixes very large menus, rather than the scrollbars that Windows uses, which don't work when dragging the mouse, the menu is instead very tall and moves to show new areas when the mouse is at the edge. My experiments seem to indicate that this works, but the exact autoscroll behavior needs tuning by doing testing with novice users.
I don't think key stroke click to get a menu is user friendly
I think you may be confused about what I meant for how to make a menubar work in a point-to-type system. What I meant was that the keystroke is handled normally by the appliation (ie if you type 'a' it inserts 'a' or whatever it should do) but that is also used as the signal that the menubar should switch to showing that application.
One thing nice that I remember from autocad (release 12 DOS) was that the menus would remember what you last did,
That is what I propose the pop-up menu should do. It can popup with the mouse pointer pointing at the previously-selected item, even if it is in a nested submenu.
You may be right. In the 60's to fight IBM you had to build hardware. In the 90's to fight Microsoft you had to write operating systems. In the future (the 20's?) to fight RedHat or whoever is the evil monopoly, you will have to write desktops and GUI toolkits.
It is getting easier each time. Perhaps after that there will be an application monopoly where you have to write applications. Eventually you will be able to compete with the monopoly by writing web sites.
If it was read-only then the macro virus could not do anything to the Linux partition (well it can still screw up the disk hardware directly but that is unavoidable).
I actually kind of like the idea of systems being able to read other systems but not write them. Instead of trying to think of everything I need to copy to the NT disk before I reboot, I can instead go there and then copy what I want from Linux. And neither system can trash the other's disks, the worst that can happen is that they will retrieve a corrupted version of the files.
Even if it can't be a filesystem there could be a "Linux Explorer" application that allows drag & drop into Explorer and the desktop, and a command-line tool that works at least as well as the ssh tools do. However I think a filesystem is possible, NT can read NFS with third-party programs.
1. I agree that "names that mean something" is a good idea, but your suggestion that Microsoft's names "mean something" is ridiculous. Instead of "Internet Explorer", lets try "Browser" or "Web Browser"
2. Unless you consider the registry editor "graphical" I don't think you can do everything in Windows "graphically". But we do need the a solution that is superior to the typical Windows gui. One HUGE advantage of text files is that they can have comments and you can easily cut & paste items to duplicate and then modify them, and you can comment out lines. Also useful is the ability in many cases to insert "if" statements. It is unclear what the solution is, but it is obvious that the most complex computer control (writing programs) even the mighty Microsoft has not come up with anything better than ASCII files. But people do like IDE's that manage those files and react nicely to the edits you do and quickly jump to the right places in them based on outside influences, so maybe some work should be done similar to that.
3. People who say "Linux can do that and did it long before Windows" are missing the point I think. I also set this up, but it was a pain and required much searching through Google and editing files and I'm quite uncertain if I did it right. Really, if I put in two network cards, why not have this turn on automatically? Why else would I have two cards? (ps Linux detected them both so it is not a hardware or driver problem) It can be done with ZERO wizards, I was all prepared to swap the network cables if it didn't work, or perhaps it could detect which end had a DHCP server. Now I don't know much about networks, but some automatic sensible setup would be nice.
4. NTFS is undocumented and Microsoft is using this as a way to stop Linux adoption. However I have heard that NTFS write is kind of working, so this may happen. A better approach would be to add some code to Windows to read Linux partitions. Then it can be balanced: each system can copy from the other but neither can damage the other with any mistake. Such code could be installed with a Windows auto-run program on a Linux installation disk.
5. I don't get this. Huge numbers of machines where I work are "installed" by inserting a HDD. This is done for Windows, Linux, and dual-boot machines, and works exactly like you suggest.
6. Actually it may be a good idea to have abstract one-syllable names like "bin" and "etc". They can be communicated in speech unambiguously. However I would eliminate all the unnecessary hierarchy and revert back to the original Linux setup where the locations were in the root. "/usr/bin" and "/usr/local/bin" and so on would be eliminated and everything dumped in/bin. The reason for all that cruft was to put things on the correct disks, but that can be solved today with symbolic links (and could be solved even better with Plan9 union mounts, or by having things that search paths instead search a fixed directory and all subdirectories and have symbolic links handle "~" and "$foo" expansions so system links can be user-specific.
7. Last I checked it has been quite awhile since 2.4, and the next version is 2.6. I suspect if you check the developers inside Microsoft you will find they use a lot of numbers to describe each version of Windows as well. Unless they really don't recompile it except when they sell a new version...
It's good to see some other people understand why overlapping windows are useful only if you can turn off click-to-raise.
I did mean sloppy focus when I said point-to-type, which is the term used when X window managers were first developed. In my experience knocking the mouse is not a problem. However pen tablets are a problem, and not likely to be solved unless they are changed to detect that the user is really holding and moving the pen, not putting it down, and such changes are not going to happen unless point-to-type becomes popular...
Menu at the top is interesting, although I think the ideal solution is to use pop-up menus. Fitt's law can be handled by having the menu popup positioned so the previously selected item is under the mouse. With that popup windows take zero screen real estate and are faster than a menu bar.
I have also tried to figure out ways to make top menubars work with point-to-type. Most solutions have tried timeouts so that quickly moving to the menubar does not switch applications, but I don't like time-based stuff. I think a better idea is to not switch the menubar until either the mouse is clicked or a keystroke is typed to an application. Moving the mouse into the menubar should put the focus back on the window that the last keystroke or mouse click went to. Unfortunately I have not had the ability to set up and test such an interface.
Actually on XP you can hit tab multiple times and it advances through all the matches, so you can see ones other than the first one.
I will agree that it blows compared to Unix shells "complete up to the first different character". I personally use tcsh set so that first tab character adds all the characters, and if there is more than one file matched the second tab character lists all the possibilities.
I think this interface could be extremely useful in a GUI (with a popup list on the second tab). For instance file choosers could be replaced with a single text field and still be user-friendly. However again we are stuck with conventions, there is no character to do the completion (tab is taken for field navigation) so Microsoft and KDE and Apple have to come up with rather unfriendly kludges like the menus of all possible completions that are always on.
On the menus: instead of a widget anywhere on the screen, why not have the menu pop up anywhere when the right mouse button is pressed? To make it user-friendly, print the word "MENU" on the right mouse button. To make it obey Fitt's law, make the menu pop up with the cursor already pointing at the item you picked last time (or use pie menus, or use both).
Of course none of this is going to happen because developers are scared to death of being "inconsistent".
If Linux becomes dominant, I would expect an explosion of user interface improvements and inventions. Within a year both KDE and Windows will seem like crude jokes compared to what Linux will be running.
If you check around there are dozens of little attempts to replace X with something new. This indicates a desire to do it, though these small projects (like DirectFB) are not going anywhere because they need major software development work and cannot get enough people interested to: 1. add a clean Xlib emulation so all existing programs work (and perhaps Windows emulation as well). 2. Provide drivers that work at least as well as the XFree86 ones. 3. Provide a drawing library at least as powerful as all the ones that draw in local bitmaps and use Xlib to blit to the screen. The first project that actually succeeded in these would instantly replace XFree.
Say Linux was popular and on a large number of desktops. Then a commercial company could throw the resources necessary at writing such a thing. They could sell it as the new whiz-bang interface where you can still run all your old programs.
With any luck there will be more than one commercial company doing this. Then there will be competition and the innovation will go very fast. There may also be some open-source competition, which will probably produce some clever ideas, though it will always have a driver problem.
Eventually what will happen is one of these companies will come out on top and will own the desktop. At that point innovation will stop and that company will probably become as powerful and evil as Microsoft, despite the fact that the Linux kernel is free (Windows runs on Intel processors where all the details are available, so I doubt the openess of Linux will prevent evil things from running atop it). Then we will enter a third dark ages, similar to the 60's under IBM and the 90's under Microsoft.
Please name any Linux program created in the last 5 years that has a menu and does not match Microsoft's menu layout.
Don't say "Emacs", it was created years before Windows. And it is available for Windows as well, so I guess that means Windows has just as inconsistent of an interface!
People have listed some clever things in KDE such as virtual desktops and middle-mouse click to lower windows or to maximize in one direction only. However, unfortunately, a lot of really good ideas have been abandoned by KDE due to the extreme pervasiveness of Windows, to the point where any even slightly different feel is immediately considered user-unfriendly.
What ideas should KDE be doing? Here are a few:
Make Point-to-type be the default. It really is better and faster to use, but it is a bit confusing to people used to click-to-type. However if you can get anybody to use it for about 1 week they will be unable to go back (verified with several non-advanced users here using NT with the registry setting that makes point-to-type work). This could be the real difference that makes Linux obviously easier to use. But I doubt this is ever going to happen because of the brainwashing recieved from Mac and Windows and now from all Linux desktops that click-to-type is the only way.
PS: the one user-unfriendliness about point-to-type that you may have encountered is when a program decides to take the focus (like when it maps a new window). Most systems leave the focus and mouse pointer in different windows, which is confusing. Some refuse to allow the program to take focus, which is very bad for modal dialogs, and somewhat unfriendly when you double-click to launch a program. I have found that if the WM warps the pointer to the nearest edge of the newly-focuesed window that it makes it clean and user-friendly. This pretty much eliminates the only advantage I have seen in click-to-type.
Perhaps more possible, and much more subtle GUI improvement: Don't raise windows unless the user clicks on the title bar (or if the program requests the window to raise, or if a "parent" raises). Besides meaning that clicking inside a window does not raise it (something some WM's can turn off), this also means that raising a child window does not move the parent, which is an annoying bug that virtually all modern window managers (and Windows) have and you can never turn off. In any case both must be turned off by default and then it is possible to work with overlapping windows. Current systems make overlapping windows impossible to use (since any work in the "lower" one will raise it, obscuring the "upper" one such that you cannot refer to it in your work), forcing everybody to do kludges like "tiled" windows (such as every IDE uses).
The second change is quite possible, and not user-unfriendly at all. In some rather primitive tests I have found that people familiar with Windows do not notice at all that the window does not raise. First of all they almost always click on the titlebar to raise windows anyway. Second my test programs also raised the window if you clicked on any "useless" area (ie the blank area around the buttons). Most users clicked on those to change focus, since anywhere else is dangerous on Windows.
I am quite serious about this, especially the no-raise idea. I have been working on various GUI systems for 12 years and it is painfully obvious how bad the raise-on-click is hurting us. I believe that this simple change would make KDE vastly easier to use and more powerful than Windows, as soon as programs started taking advantage of it to draw larger displays and control panels.
Absolutely false. Think about it, if this was true, then anybody could steal any copyrighted work and sell copies of it, and escape prosecution by saying they licensed those copies under "those GPL rules that mean that I am not liable for intellectual property issues". The idea is totally absurd: it would immediately make copyrights useless and unenforcable as anybody could instantly make themselves "not liable" by giving or selling one copy.
What I meant was that this "problem" is not unique to the GPL. Workers at MicroSoft encounter copyrighted code in thousands of ways. Some of it belongs to big companies that are much more likely to sue than the OSS advocates are. Yet they manage to continue with a few rules like "don't copy this, and if you are caught copying it then you are in a heap of trouble!".
Saying that the GPL is somehow special or more dangerous than the thousands and thousands of other copyrights and contracts that exists in the programmng business is pure FUD. Anybody saying anything about having to be "careful" about GPL code without also saying that that is true about virtually every non-GPL piece of code is spreading it.
Slashdot Mirror
What you say true that if you assumme you can make a profit from selling GPL code (by either selling support, or copyrighted addon's, or just encouraging customers to buy the original version rather than a copy). TrollTech does assumme this, and yes they had to modify the GPL to their license that disallows redistribution for any profit, even of a GPL version.
However the vast majority of people, especially everybody who does not like the GPL, claim that it is a fact that you cannot make a profit selling GPL code (even though I kind of like the GPL, I can agree with this somewhat, people are cheapskates and if the free copy is legal they will get it). Therefore if you make this assumption you can assume that only the original author can profit from the GPL code, since only they can dual-license it into a form that you can make a profit from.
It is refreshing to see that even an anti-GPL advocate can admit you can make money using the GPL, however. You seem more convinced of it than I am.
A more practical response is that all theoretical damage of the GPL is well known and is avoided in the real world. All code of any value for developing new and innovate programs is released using licenses like the LGPL, LGPL with exceptions to make it even more like public-domain, or BSD or public-domain licenses.
I write LGPL+exceptions code (the exceptions are the typical ones where you can use the library as though it was a public-domain closed source library, but if you modify any of the source code to the library you must release all the changes if you redistribute it). I know for a fact that my code would not be used if it was GPL so I am strongly encouraged to not do that. This is true of almost any writer of open-source code that is designed to be a service, and thus completely defuses any argument that it somehow harms advancement and use by anybody. I also write GPL code but that is because it is an end application, that code is useless to anybody except to make a very similar program, which I would consider the theft of my ideas and IP. Also I know that the fact that it is GPL has zero impact on the popularity of this code, because it has zero impact on the ability to run it, which is the main use.
Check how TrollTech is doing and their business model for details.
Microsoft has released quite a lot of stuff (example code) public domain, which is a FSF-compatable license.
They also make available all the code to the GPL tools they sell with their Unix-compatability package, that code is real GPL.
Though I don't know if they modified anything, so perhaps that isn't really contributing any actual work they did under the GPL.
The reason that that GPL does not force you to release changes to the code if you don't distribute it is practical:
The GPL is designed entirely to be an "exception" to normal copyrights. If you ignore or "disobey" the GPL then all that is left is normal copyright laws, which lets you do *less* than the GPL allows. Therefore there is zero reason to "disobey" the GPL because you gain nothing and you lose a slight amount (perhaps nothing if you did not plan to redistribute anyway).
Because the GPL does not restrict you from doing anything you normally could do with copyrighted code, it cannot be considered a "contract". Thus it cannot be challenged with any "I didn't agree to that" arguments that make EULA and click-to-agree type contracts legally useless.
Now normally when you buy a book, you are allowed to scribble all over it for any reason you want. Copyrights and other laws do not prevent that. So the GPL has to allow what copyright does, so in no way can it be considered a restriction.
Now to be fair, the GPL could say that you must give RMS your first-born. If you chose to disobey that part of the GPL, and a court agreed with you, you would still be bound by copyright and unable to copy the GPL code, but you would not have to give up your first-born. So adding restrictions would in theory not harm the GPL because legally at worst it would end up where it is now. However in the court of public opinion, having *any* part of it challenged would cause people to think the whole thing is meaningless.
This is also why "advertising" clauses are not allowed by the GPL, and why there are no restrictions allowed on what the software is used for.
I was going to say this is unlikely, but in fact it is quite possible that Sun did everything legally, and for the benifit of all parties. AT&T got money for the rights and may still retain the rights for use of the code in non-GPL products. Sun paid money for the rights to use the code in their products, and by adding to the Linux code insured compatability and extensive testing of this code, and avoided the expense of supporting two different code sets. Most likely both AT&T and Sun got some other code that is valuable that did not go into Linux. And of course Linux got some useful code.
This leads to the obvious conclusion that Microsoft is funding them. And they have made a big point that Microsoft bought a license from them at an apparently very inflated price (many times what Sun paid apparently, does anybody know how much Sun paid?)
However there is another possibility: SCO is making people believe that Microsoft is paying them. They want people to believe they have this new business plan where they will collect money from Microsoft as payment for making press releases, and this is a stable source of huge profits. In some ways it makes a lot of sense: if Microsoft is paying them, money invested in SCO stock a month ago and sold now would collect you a whole lot more of the profits of Microsoft than if you had invested in Microsoft stock directly.
I seriously belive that most, if not all, the outside investors who are buying their stock belive this, and are hoping for Microsoft to purchase more licenses from SCO.
The fact that they will not reveal any of the code proves that they have no case and their purpose is to generate FUD about Linux.
Some of SCO's actions seem even designed to help investors believe this. Most of the press releases clearly claim "Linux is crap" which is pretty foolish if you actually want people to pay you for it rather than go with something else.
Absolutely. The problem is that they don't believe it.
But if they provided cygwin-like functionality by default in Windows they would eliminate 50% of the reasons people work with Linux. If they made it easy to port skills back and forth amoung the systems they would get rid of a huge roadblock to integrating Windows into Unix environments and to assimilating people with Unix skills. And they can compete in a *nice* way: add a "feature" that is useful, but not vital, so portable programs can ifdef it out, and just be less fancy. Document that feature, or even better provide a toy version that runs on Linux, so users can be pretty certain that the interface for that feature is defined and may be portable in the future. Doing this will kill any incentive for Linux programmers to make a different implementation, so the Windows one will be obviously better. This would be successful competition, but will also completely defuse the hate most programmers (like me) have towards Microsofts actions.
Linux advocates don't like to admit it, but if NT had appeared in 1990 with cygwin (or even Xenix) functionality, Linux would be nowhere today and Microsoft would have the total monopoly that they wanted. Oddly enough though, Microsoft refuses to admit this either, somewhere deep in Bill Gates and the NT designer's psyches they know they blew it by ignoring standards, but they are loathe to admit it, even though that would be the smartest thing they could do for their company.
Vertical scrolling is currently done by moving the entire menu window, not it's contents. This happens when the mouse is moved to the edge of the screen where the menu goes off the screen. This does have some problems, for instance to stop the scrolling the user must move the mouse at least 1 pixel away from the edge. Horizontal scrolling is not implemented yet. I think the scrolling behavior I have needs work and testing, also X is not really all that hot at moving windows, for instance I only move one window, not the whole hierarchy, because it won't move them nicely.
I have not figured out any way to handle the popup menu whereever the mouse is without placing some of them off-screen. Moving the mouse pointer is bad UI, I tried it, and tried moving it back after release. This was not very user-friendly.
You are the idiot for thinking that anybody who suggests anything different than the Microsoft line is not being imaginative.
And I will stand by my suggestion that short abstract names are better. They do not convey meaning that can confuse people about the real purpose, and they are much easier to type and to communicate vocally. Does "Program Names" have a space in it or not? Why not call it "Programs". And why are there things in there that are not programs? And what do French people think about that name? In my experience *nobody* misspells "bin" and it can be clearly stated over the phone and in email.
This however is the reason behind most of the inconsistency of Linux (and Windows!) applications. Why doesn't ctrl+s save in all programs? Mostly because a lot of programs assigned ctrl+s to do something else and people don't want to change the interface.
I personally believe the mouse should have a dedicated menu key, with the word "Menu" printed on it (or a little picture, but then the menus cannot change to pies or other shapes). And programs should not use it for anything other than menus.
Position of menus when near edge of screen
My solution to this is to pop it up in the correct position, with it going off the screen, and then have autoscroll that moves it in if you move the mouse to the edge. This also fixes very large menus, rather than the scrollbars that Windows uses, which don't work when dragging the mouse, the menu is instead very tall and moves to show new areas when the mouse is at the edge. My experiments seem to indicate that this works, but the exact autoscroll behavior needs tuning by doing testing with novice users.
I don't think key stroke click to get a menu is user friendly
I think you may be confused about what I meant for how to make a menubar work in a point-to-type system. What I meant was that the keystroke is handled normally by the appliation (ie if you type 'a' it inserts 'a' or whatever it should do) but that is also used as the signal that the menubar should switch to showing that application.
One thing nice that I remember from autocad (release 12 DOS) was that the menus would remember what you last did,
That is what I propose the pop-up menu should do. It can popup with the mouse pointer pointing at the previously-selected item, even if it is in a nested submenu.
It is getting easier each time. Perhaps after that there will be an application monopoly where you have to write applications. Eventually you will be able to compete with the monopoly by writing web sites.
I actually kind of like the idea of systems being able to read other systems but not write them. Instead of trying to think of everything I need to copy to the NT disk before I reboot, I can instead go there and then copy what I want from Linux. And neither system can trash the other's disks, the worst that can happen is that they will retrieve a corrupted version of the files.
Even if it can't be a filesystem there could be a "Linux Explorer" application that allows drag & drop into Explorer and the desktop, and a command-line tool that works at least as well as the ssh tools do. However I think a filesystem is possible, NT can read NFS with third-party programs.
1. I agree that "names that mean something" is a good idea, but your suggestion that Microsoft's names "mean something" is ridiculous. Instead of "Internet Explorer", lets try "Browser" or "Web Browser"
/bin. The reason for all that cruft was to put things on the correct disks, but that can be solved today with symbolic links (and could be solved even better with Plan9 union mounts, or by having things that search paths instead search a fixed directory and all subdirectories and have symbolic links handle "~" and "$foo" expansions so system links can be user-specific.
2. Unless you consider the registry editor "graphical" I don't think you can do everything in Windows "graphically". But we do need the a solution that is superior to the typical Windows gui. One HUGE advantage of text files is that they can have comments and you can easily cut & paste items to duplicate and then modify them, and you can comment out lines. Also useful is the ability in many cases to insert "if" statements. It is unclear what the solution is, but it is obvious that the most complex computer control (writing programs) even the mighty Microsoft has not come up with anything better than ASCII files. But people do like IDE's that manage those files and react nicely to the edits you do and quickly jump to the right places in them based on outside influences, so maybe some work should be done similar to that.
3. People who say "Linux can do that and did it long before Windows" are missing the point I think. I also set this up, but it was a pain and required much searching through Google and editing files and I'm quite uncertain if I did it right. Really, if I put in two network cards, why not have this turn on automatically? Why else would I have two cards? (ps Linux detected them both so it is not a hardware or driver problem) It can be done with ZERO wizards, I was all prepared to swap the network cables if it didn't work, or perhaps it could detect which end had a DHCP server. Now I don't know much about networks, but some automatic sensible setup would be nice.
4. NTFS is undocumented and Microsoft is using this as a way to stop Linux adoption. However I have heard that NTFS write is kind of working, so this may happen. A better approach would be to add some code to Windows to read Linux partitions. Then it can be balanced: each system can copy from the other but neither can damage the other with any mistake. Such code could be installed with a Windows auto-run program on a Linux installation disk.
5. I don't get this. Huge numbers of machines where I work are "installed" by inserting a HDD. This is done for Windows, Linux, and dual-boot machines, and works exactly like you suggest.
6. Actually it may be a good idea to have abstract one-syllable names like "bin" and "etc". They can be communicated in speech unambiguously. However I would eliminate all the unnecessary hierarchy and revert back to the original Linux setup where the locations were in the root. "/usr/bin" and "/usr/local/bin" and so on would be eliminated and everything dumped in
7. Last I checked it has been quite awhile since 2.4, and the next version is 2.6. I suspect if you check the developers inside Microsoft you will find they use a lot of numbers to describe each version of Windows as well. Unless they really don't recompile it except when they sell a new version...
I did mean sloppy focus when I said point-to-type, which is the term used when X window managers were first developed. In my experience knocking the mouse is not a problem. However pen tablets are a problem, and not likely to be solved unless they are changed to detect that the user is really holding and moving the pen, not putting it down, and such changes are not going to happen unless point-to-type becomes popular...
Menu at the top is interesting, although I think the ideal solution is to use pop-up menus. Fitt's law can be handled by having the menu popup positioned so the previously selected item is under the mouse. With that popup windows take zero screen real estate and are faster than a menu bar.
I have also tried to figure out ways to make top menubars work with point-to-type. Most solutions have tried timeouts so that quickly moving to the menubar does not switch applications, but I don't like time-based stuff. I think a better idea is to not switch the menubar until either the mouse is clicked or a keystroke is typed to an application. Moving the mouse into the menubar should put the focus back on the window that the last keystroke or mouse click went to. Unfortunately I have not had the ability to set up and test such an interface.
I will agree that it blows compared to Unix shells "complete up to the first different character". I personally use tcsh set so that first tab character adds all the characters, and if there is more than one file matched the second tab character lists all the possibilities.
I think this interface could be extremely useful in a GUI (with a popup list on the second tab). For instance file choosers could be replaced with a single text field and still be user-friendly. However again we are stuck with conventions, there is no character to do the completion (tab is taken for field navigation) so Microsoft and KDE and Apple have to come up with rather unfriendly kludges like the menus of all possible completions that are always on.
Of course none of this is going to happen because developers are scared to death of being "inconsistent".
If you check around there are dozens of little attempts to replace X with something new. This indicates a desire to do it, though these small projects (like DirectFB) are not going anywhere because they need major software development work and cannot get enough people interested to: 1. add a clean Xlib emulation so all existing programs work (and perhaps Windows emulation as well). 2. Provide drivers that work at least as well as the XFree86 ones. 3. Provide a drawing library at least as powerful as all the ones that draw in local bitmaps and use Xlib to blit to the screen. The first project that actually succeeded in these would instantly replace XFree.
Say Linux was popular and on a large number of desktops. Then a commercial company could throw the resources necessary at writing such a thing. They could sell it as the new whiz-bang interface where you can still run all your old programs.
With any luck there will be more than one commercial company doing this. Then there will be competition and the innovation will go very fast. There may also be some open-source competition, which will probably produce some clever ideas, though it will always have a driver problem.
Eventually what will happen is one of these companies will come out on top and will own the desktop. At that point innovation will stop and that company will probably become as powerful and evil as Microsoft, despite the fact that the Linux kernel is free (Windows runs on Intel processors where all the details are available, so I doubt the openess of Linux will prevent evil things from running atop it). Then we will enter a third dark ages, similar to the 60's under IBM and the 90's under Microsoft.
Don't say "Emacs", it was created years before Windows. And it is available for Windows as well, so I guess that means Windows has just as inconsistent of an interface!
What ideas should KDE be doing? Here are a few:
Make Point-to-type be the default. It really is better and faster to use, but it is a bit confusing to people used to click-to-type. However if you can get anybody to use it for about 1 week they will be unable to go back (verified with several non-advanced users here using NT with the registry setting that makes point-to-type work). This could be the real difference that makes Linux obviously easier to use. But I doubt this is ever going to happen because of the brainwashing recieved from Mac and Windows and now from all Linux desktops that click-to-type is the only way.
PS: the one user-unfriendliness about point-to-type that you may have encountered is when a program decides to take the focus (like when it maps a new window). Most systems leave the focus and mouse pointer in different windows, which is confusing. Some refuse to allow the program to take focus, which is very bad for modal dialogs, and somewhat unfriendly when you double-click to launch a program. I have found that if the WM warps the pointer to the nearest edge of the newly-focuesed window that it makes it clean and user-friendly. This pretty much eliminates the only advantage I have seen in click-to-type.
Perhaps more possible, and much more subtle GUI improvement: Don't raise windows unless the user clicks on the title bar (or if the program requests the window to raise, or if a "parent" raises). Besides meaning that clicking inside a window does not raise it (something some WM's can turn off), this also means that raising a child window does not move the parent, which is an annoying bug that virtually all modern window managers (and Windows) have and you can never turn off. In any case both must be turned off by default and then it is possible to work with overlapping windows. Current systems make overlapping windows impossible to use (since any work in the "lower" one will raise it, obscuring the "upper" one such that you cannot refer to it in your work), forcing everybody to do kludges like "tiled" windows (such as every IDE uses).
The second change is quite possible, and not user-unfriendly at all. In some rather primitive tests I have found that people familiar with Windows do not notice at all that the window does not raise. First of all they almost always click on the titlebar to raise windows anyway. Second my test programs also raised the window if you clicked on any "useless" area (ie the blank area around the buttons). Most users clicked on those to change focus, since anywhere else is dangerous on Windows.
I am quite serious about this, especially the no-raise idea. I have been working on various GUI systems for 12 years and it is painfully obvious how bad the raise-on-click is hurting us. I believe that this simple change would make KDE vastly easier to use and more powerful than Windows, as soon as programs started taking advantage of it to draw larger displays and control panels.
Absolutely false. Think about it, if this was true, then anybody could steal any copyrighted work and sell copies of it, and escape prosecution by saying they licensed those copies under "those GPL rules that mean that I am not liable for intellectual property issues". The idea is totally absurd: it would immediately make copyrights useless and unenforcable as anybody could instantly make themselves "not liable" by giving or selling one copy.
IBM is now the Microsoft of Linux, at least according to a common opinion presented here that IBM is funding little RedHat's lawsuit.
Saying that the GPL is somehow special or more dangerous than the thousands and thousands of other copyrights and contracts that exists in the programmng business is pure FUD. Anybody saying anything about having to be "careful" about GPL code without also saying that that is true about virtually every non-GPL piece of code is spreading it.