Re:Two schools of thought
on
Think Unix
·
· Score: 1
"The other approach is the Unix one, which says that it's the user's responsibility to learn the system, warts and all. This is exactly the mentality that created the "Unix wizard" syndrome in the first place (people have to work hard to learn how to use the system effectively, and then they don't want it simplified for others, which would only devalue their accomplishment), and sadly, there seems to be little going on in the Linux world to counteract that."
If there is so little going on in the Linux world to counteract the Unix wizard mentality, what do you call KDE and GNOME? Both projects are about making the Unix desktop easier to use, and while neither have fully succeeded on that goal yet, both projects are going strong and are high-profile in the Linux world, hardly what I'd call "little going on".
"I think good UI is intuitable, i.e. users can easily use intuition to figure out how to accomplish a task."
The problem is that inituition can only go so far. At some point, one actually has to figure out the basics of how things work, such as moving, and copying files, what a program is, customizing users settings, etc. These things really can't be learned just by poking around. They have to be taught. The trick is not to make it so that the user doesn't have to be taught anything, but rather that the user has to learn a relatively small number of basic things in order to be productive.
As far as I can tell, the issues that Stallman brought up deal with 1) the GPL rights of those who had copies of KDE made before Qt 2.2 was dual licensed under the QPL and GPL, and 2) the GPL rights of KDE developers who incorporated third-party GPL'd code into software linked with Qt. Neither of those issues affects the current KDE2 codebase, so Debian is in the clear.
The fight is *mostly* over, but KDE has some leftover stuff to clean up.
RMS has a point about getting "forgiveness" or permission from the authors of third-party GPL code that made it into KDE. If the QPL and GPL did conflict, then the KDE developers did breach the GPL of that third-party code, and since one's rights under the GPL end when one breaches its terms, the KDE developers who introduced the third-party code no longer have a right to touch that code, period. Troll Tech licensing Qt 2.2 under the GPL can't undo the breach that already took place. The only way for those KDE developers to get their GPL rights back on that third-party code is to get permission from the copyright holders of the third-party code.
RMS was undiplomatic, but he was unfortunately correct.
I believe that some pieces of Mozilla already were dual-licensed. Probably the reason that the Mozilla crew is seeing to dual-license all of Mozilla is because of problems with Galeon's licensing and probably also to head off any possible licensing controversies with Nautilus embedding Mozilla as well.
"KDE is much closer to the current CDE in the look and feel area."
That's only true for KDE 1.x, not KDE 2, and even then KDE 1.x and CDE don't look *that* much alike.
The politics of open source licensing also may have something to do with it. GNOME, whatever its faults, is definitely free software, no question about it. KDE has been dogged by licensing controversies in one form or another since day one, and Sun probably just does not want to get drawn into that mess.
"If I want to use some little GNOME program (say, Galeon), what do I need to do?"
"Download the program.
Figure out which libraries are needed for the GNOME stuff.
Figure out which libraries are needed BY the GNOME stuff.
Locate and download all those libraries.
Find a place to put all those libraries.
Debug all my existing applications because I just upgraded all my libraries (can you say "DLL hell"?)
Occasionally: Answer "NO" to a program that wants to "associate files of type ABC with this program"
Funny, all I've usually ever needed to do is download and install the program. If one already has GNOME installed, all the libraries that a GNOME program depends on are already there. I've never had problems with debugging apps because I just upgraded libraries.
Galeon is a poor example of a generic GNOME program. It is new and beta, and it depends upon Mozilla components as well as the GNOME libraries.
Offhand, you seem to be objecting to problems that for the most part don't exist.
"Why should I be stuck with gnome or kde or twm when I can switch at will?"
I think you missed the point entirely. This has very little to do with GUIs per se. What Miguel is trying to implement is a free (as opposed to proprietary) architecture to build applications from building blocks, using CORBA more or less as the glue that binds these blocks together. You would not have to run the GNOME desktop to get the benefit of the component technologies that Miguel was talking about. All you would need would be the appropriate libs and daemons running.
"Even though it is nice that X11 is made so extensible, this still means that a standard X11 setup uses none of these extentions"
That depends on what you mean by a "standard X11 setup." AFAIK, the "Shape" extension to X11 is pretty darn common, so in effect that would be a de facto standard part of an X setup. In short, a so-called "standard" X setup could very well actually be plain X + several common extensions.
. . . was that the text under the icons on the desktop wasn't highlighted in some colored rectangle, with the exception of the text "HOME" under what is presumably the icon for the home directory. From my experience with trying out the KDE2 betas, this looks like a bad move, because the text under the icons could become very hard to read if someone chooses a fancy background from, say, Propanganda or Digital Blasphemy. Highlighted text under icons is always readable no matter what the background. I hope the Nautilus folks have considered this.
"This boild down to your rights vs my rights, (the basic version) we both have rights to do pretty much whatever we please AS LONG AS it does not infringe others' rights to do whatever they please."
Very true. This still presumes that infringing on another's rights is wrong--and issues of right and wrong are otherwise known as moral issues. This is why I said that it is not a matter of whether government should legislate morality at all, but rather how much morality should be enforced by the legal system.
"The point is, the government should not, and can not make moral decisions for us."
There is quite a bit of truth in the maxim "You can't legislate morality," but the issue gets a bit more complicated. For example, there are laws against fraud, rape, robbery, and murder. There is an implicit assumption involved that those things are made illegal because they are morally wrong. One can argue that the basis of these laws is that they constitute an aggression of one party onto someone else, but even this presumes that aggression against another is wrong. The question really isn't "Should government legislate morality?" but rather "What parts of morality should be legislated, and what should be left to free choice?"
The best argument against censorship is that government cannot be fully trusted, and that giving government the power to censor means giving it the power to cover up and quash dissenting opinions that could expose failure and/or wrongdoing on the part of the government. This argument has little to do with not legislating morality, but rather it presumes that power corrupts, and that government would use the power to censor for immoral purposes.
"DLL Hell -.SO Hell - how many times you downloaded binary package just to find out it needs 4.523.2 version of particular library ?"
Very seldom, actually, and when I have had that problem it has generally been with software under heavy development (i.e. GNOME back when the latest GTK+ was at version 1.1.x)
"You install it and then all the other programs break."
Um, no. If I install it in spite of unsatisfied dependencies, the programs in the package won't work, but the rest of the system won't break.
Also note that binary packages for Linux don't generally install their own versions of system libraries, and package managers cry foul at binary packages that do try to mess with system libraries. Binary packages for Windows often do install their own versions of system libs, and up until recently, Windows didn't complain. That's what causes the DLL Hell.
"A *n?x box of some kind seems to be the best choice for this . . . "
"I realize this is slashdot, where all things MS are evil, but c'mon - the majority of people have to deal with Windows... teach them on the most prevalent platform"
Windows may be the most prevalent platform for people *using* computers, but that doesn't mean that it is the best system for teaching people how to understand computers. Windows is designed so that its complexity and workings are hidden. You *can* peek under the hood to some extent, but it was not by and large put together to allow poking, prodding, and exploration. Unix, on the other hand, was not designed to hide its complexity, and attempts to do so, such as the CDE, KDE, and GNOME, came well after its inception. Config files are plain text. Things that are presented as integrated monoliths in Windows and the Mac are far more separated in Unix; the shell, the login, the low-level basis for the GUI (X), and the GUI widgets, are all separate programs or libraries. It's easier to poke around and figure out what does what.
No, as opposed to a newer, bugfixed, and rather stable GNOME. While GNOME 1.0 was definitely core dump city, GNOME has been pretty solid ever since October GNOME. I know because I've used both October GNOME and GNOME 1.2 from Helix Code.
I *don't* hate KDE, but I suspect one of the reasons for hating it has probably been the various licensing issues with Qt that have dogged KDE almost since its inception and provided the fodder for many a flame war.
"Keep in mind that the Unix security model, which is damned good by the way, assumes that the machine's owner (root, anyway) is God. This is not a bad thing."
Actually, having an account that has total control of the system (the superuser) is probably the main weakness of Unix security, because it means that there is a single point of failure; if root is compromised, the whole machine is compromised. A finer-grained method of establishing who has what privileges, a.k.a. capabilities, is more secure.
An open CDE might be a "spoiler" in the competition(?) between KDE and GNOME, an option for those dissatisfied with either of those two desktops, or just with the flamewars erupting over those desktops. Also, CDE is in some sense already a "standard" desktop, albeit one that is not widely on Linux for the time being. That may be comforting to some.
Take a look at the current poll on www.motifzone.net. The project about 70% of the people who responded to the poll wanted was Open CDE.
"The problem is that NT labor is readily available, while Unix/Linux labor is not."
Perhaps, but if one already is a shop running SCO in the first place, then one has probably already hired people to admin the SCO boxes, and those would be people who'd be skilled in Unix. The Unix/Linux labor (or at least Unix labor) that one would need would already be on hand.
Slashdot Mirror
"The other approach is the Unix one, which says that it's the user's responsibility to learn the system, warts and all. This is exactly the mentality that created the "Unix wizard" syndrome in the first place (people have to work hard to learn how to use the system effectively, and then they don't want it simplified for others, which would only devalue their accomplishment), and sadly, there seems to be little going on in the Linux world to counteract that."
If there is so little going on in the Linux world to counteract the Unix wizard mentality, what do you call KDE and GNOME? Both projects are about making the Unix desktop easier to use, and while neither have fully succeeded on that goal yet, both projects are going strong and are high-profile in the Linux world, hardly what I'd call "little going on".
"I think good UI is intuitable, i.e. users can easily use intuition to figure out how to accomplish a task."
The problem is that inituition can only go so far. At some point, one actually has to figure out the basics of how things work, such as moving, and copying files, what a program is, customizing users settings, etc. These things really can't be learned just by poking around. They have to be taught. The trick is not to make it so that the user doesn't have to be taught anything, but rather that the user has to learn a relatively small number of basic things in order to be productive.
I invite you to read about an "experiment" that a physicist did to show the problems with postmodern thought, including the issue that you brought up.
As far as I can tell, the issues that Stallman brought up deal with 1) the GPL rights of those who had copies of KDE made before Qt 2.2 was dual licensed under the QPL and GPL, and 2) the GPL rights of KDE developers who incorporated third-party GPL'd code into software linked with Qt. Neither of those issues affects the current KDE2 codebase, so Debian is in the clear.
The fight is *mostly* over, but KDE has some leftover stuff to clean up.
RMS has a point about getting "forgiveness" or permission from the authors of third-party GPL code that made it into KDE. If the QPL and GPL did conflict, then the KDE developers did breach the GPL of that third-party code, and since one's rights under the GPL end when one breaches its terms, the KDE developers who introduced the third-party code no longer have a right to touch that code, period. Troll Tech licensing Qt 2.2 under the GPL can't undo the breach that already took place. The only way for those KDE developers to get their GPL rights back on that third-party code is to get permission from the copyright holders of the third-party code.
RMS was undiplomatic, but he was unfortunately correct.
I didn't know that Troll Tech was getting help from RMS. Where did you here about that part?
I believe that some pieces of Mozilla already were dual-licensed. Probably the reason that the Mozilla crew is seeing to dual-license all of Mozilla is because of problems with Galeon's licensing and probably also to head off any possible licensing controversies with Nautilus embedding Mozilla as well.
"KDE is much closer to the current CDE in the look and feel area."
That's only true for KDE 1.x, not KDE 2, and even then KDE 1.x and CDE don't look *that* much alike.
The politics of open source licensing also may have something to do with it. GNOME, whatever its faults, is definitely free software, no question about it. KDE has been dogged by licensing controversies in one form or another since day one, and Sun probably just does not want to get drawn into that mess.
"If I want to use some little GNOME program (say, Galeon), what do I need to do?"
"Download the program.
Figure out which libraries are needed for the GNOME stuff.
Figure out which libraries are needed BY the GNOME stuff.
Locate and download all those libraries.
Find a place to put all those libraries.
Debug all my existing applications because I just upgraded all my libraries (can you say "DLL hell"?)
Occasionally: Answer "NO" to a program that wants to "associate files of type ABC with this program"
Funny, all I've usually ever needed to do is download and install the program. If one already has GNOME installed, all the libraries that a GNOME program depends on are already there. I've never had problems with debugging apps because I just upgraded libraries.
Galeon is a poor example of a generic GNOME program. It is new and beta, and it depends upon Mozilla components as well as the GNOME libraries.
Offhand, you seem to be objecting to problems that for the most part don't exist.
"Why should I be stuck with gnome or kde or twm when I can switch at will?"
I think you missed the point entirely. This has very little to do with GUIs per se. What Miguel is trying to implement is a free (as opposed to proprietary) architecture to build applications from building blocks, using CORBA more or less as the glue that binds these blocks together. You would not have to run the GNOME desktop to get the benefit of the component technologies that Miguel was talking about. All you would need would be the appropriate libs and daemons running.
"Even though it is nice that X11 is made so extensible, this still means that a standard X11 setup uses none of these extentions"
That depends on what you mean by a "standard X11 setup." AFAIK, the "Shape" extension to X11 is pretty darn common, so in effect that would be a de facto standard part of an X setup. In short, a so-called "standard" X setup could very well actually be plain X + several common extensions.
I've tried to get EFM from CVS, but all I get is:
cvs [login aborted]: recv() from server cvs.enlightenment.org: Connection reset by peer
Wah. Wish that there were daily snaps of EFM. It looks rather interesting.
. . . was that the text under the icons on the desktop wasn't highlighted in some colored rectangle, with the exception of the text "HOME" under what is presumably the icon for the home directory. From my experience with trying out the KDE2 betas, this looks like a bad move, because the text under the icons could become very hard to read if someone chooses a fancy background from, say, Propanganda or Digital Blasphemy. Highlighted text under icons is always readable no matter what the background. I hope the Nautilus folks have considered this.
"This boild down to your rights vs my rights, (the basic version) we both have rights to do pretty much whatever we please AS LONG AS it does not infringe others' rights to do whatever they please."
Very true. This still presumes that infringing on another's rights is wrong--and issues of right and wrong are otherwise known as moral issues. This is why I said that it is not a matter of whether government should legislate morality at all, but rather how much morality should be enforced by the legal system.
"The point is, the government should not, and can not make moral decisions for us."
There is quite a bit of truth in the maxim "You can't legislate morality," but the issue gets a bit more complicated. For example, there are laws against fraud, rape, robbery, and murder. There is an implicit assumption involved that those things are made illegal because they are morally wrong. One can argue that the basis of these laws is that they constitute an aggression of one party onto someone else, but even this presumes that aggression against another is wrong. The question really isn't "Should government legislate morality?" but rather "What parts of morality should be legislated, and what should be left to free choice?"
The best argument against censorship is that government cannot be fully trusted, and that giving government the power to censor means giving it the power to cover up and quash dissenting opinions that could expose failure and/or wrongdoing on the part of the government. This argument has little to do with not legislating morality, but rather it presumes that power corrupts, and that government would use the power to censor for immoral purposes.
Go to the Computer Songs and Parodies site.
"DLL Hell - .SO Hell - how many times you downloaded binary package just to find out it needs 4.523.2 version of particular library ?"
Very seldom, actually, and when I have had that problem it has generally been with software under heavy development (i.e. GNOME back when the latest GTK+ was at version 1.1.x)
"You install it and then all the other programs break."
Um, no. If I install it in spite of unsatisfied dependencies, the programs in the package won't work, but the rest of the system won't break.
Also note that binary packages for Linux don't generally install their own versions of system libraries, and package managers cry foul at binary packages that do try to mess with system libraries. Binary packages for Windows often do install their own versions of system libs, and up until recently, Windows didn't complain. That's what causes the DLL Hell.
"I like using the year as the version number.
"I think it makes sense to have product version YYYY.MM.DD"
Unfortunately, that method of version numbering tends to be used for snapshots and unstable code in development.
"A *n?x box of some kind seems to be the best choice for this . . . "
"I realize this is slashdot, where all things MS are evil, but c'mon - the majority of people have to deal with Windows... teach them on the most prevalent platform"
Windows may be the most prevalent platform for people *using* computers, but that doesn't mean that it is the best system for teaching people how to understand computers. Windows is designed so that its complexity and workings are hidden. You *can* peek under the hood to some extent, but it was not by and large put together to allow poking, prodding, and exploration. Unix, on the other hand, was not designed to hide its complexity, and attempts to do so, such as the CDE, KDE, and GNOME, came well after its inception. Config files are plain text. Things that are presented as integrated monoliths in Windows and the Mac are far more separated in Unix; the shell, the login, the low-level basis for the GUI (X), and the GUI widgets, are all separate programs or libraries. It's easier to poke around and figure out what does what.
If it has enough features so that people can make interesting stuff with it, then a lot of people will probably be interested.
No, as opposed to a newer, bugfixed, and rather stable GNOME. While GNOME 1.0 was definitely core dump city, GNOME has been pretty solid ever since October GNOME. I know because I've used both October GNOME and GNOME 1.2 from Helix Code.
I suspect I'm feeding a troll, though. Sigh.
I *don't* hate KDE, but I suspect one of the reasons for hating it has probably been the various licensing issues with Qt that have dogged KDE almost since its inception and provided the fodder for many a flame war.
"Keep in mind that the Unix security model, which is damned good by the way, assumes that the machine's owner (root, anyway) is God. This is not a bad thing."
Actually, having an account that has total control of the system (the superuser) is probably the main weakness of Unix security, because it means that there is a single point of failure; if root is compromised, the whole machine is compromised. A finer-grained method of establishing who has what privileges, a.k.a. capabilities, is more secure.
An open CDE might be a "spoiler" in the competition(?) between KDE and GNOME, an option for those dissatisfied with either of those two desktops, or just with the flamewars erupting over those desktops. Also, CDE is in some sense already a "standard" desktop, albeit one that is not widely on Linux for the time being. That may be comforting to some.
Take a look at the current poll on www.motifzone.net. The project about 70% of the people who responded to the poll wanted was Open CDE.
"The problem is that NT labor is readily available, while Unix/Linux labor is not."
Perhaps, but if one already is a shop running SCO in the first place, then one has probably already hired people to admin the SCO boxes, and those would be people who'd be skilled in Unix. The Unix/Linux labor (or at least Unix labor) that one would need would already be on hand.