The biggest problem that I face with Postgres 7.4 is that its referential integrity's locks block INSERTS and UPDATES that should go through fine.
For example, set up these two tables: CREATE TABLE car_type (
id serial primary key,
name varchar(20)
);
CREATE TABLE car (
id serial primary key,
car_type_id integer references car_type );
Now, try having two different connections insert with the same foreign key value (this field does not have a unique constraint):
Connection1: BEGIN; INSERT INTO car (car_type_id) values(1);
Connection2: BEGIN; INSERT INTO car (car_type_id) values(1);
You will see that the second transaction is waiting for the first transaction to commit. That is just rediculous and is one reason that Postgres is still small time.
With iTunes, Final Cut Pro, and iMac, Apple obviously sees apps and services instead of the OS as the key making money off of its expertise. Even Microsoft is not secure in its OS market share and is trying to lock in users with Passport and.NET. The only downside is that the user interfaces between OSes will become even more homegenous once the OS doesn't matter.
Re:This book is destined to become a classic
on
Programming PHP
·
· Score: 1
PHP is a widely used embedded-in-html scripting language, but it is by no means the first. PHP started off as a templating language, but features were continually added which it caused it to grow beyond Server Side Includes. As such, PHP is not a well designed scripting language. If you separate your code into functions, it is extremely difficult to track down errors, because you only get the line that the error occured on and not the line that called that function.
PHP is definitely very easy to start programming dynamic web pages, but it is increasingly problematic as your code base grows. If you primarily do web design and just want a little dynamic content, PHP's shortcomings won't affect you. However, web based applications will run into problems with PHP repeatedly. Besides the problem where it copies objects out of an array when you use a "foreach", you are also stuck remembering to put an amperstand in front of every object so it gets passed or assigned by reference and not value. For example:
$x = $obj; # $x is a copy of $obj
$x = &$obj; # $x and $obj refer to the same object
Besides the overhead of copying large objects, it just creates one more "gotcha" that you have to remember to avoid. It is not that good code can't be written, but it is so much easier to write problematic code.
There are a lot of differences in the model where the CueCat is free and use of it is free but in theory you will buy more stuff through CueCat, therefore the companies that use it will be willing to pay for it. Since the CueCat is hardware, there are some inescapable costs of production and distribution that DigitalConvergence can only reasonably make up for if a certain percentage will actually use it with their software. Geeks that develop alternative software will have a reason to get as many free cuecats as possible and DigitalConvergence won't benefit from it. If just a few people were re-purposing the cuecat for their own use, it still wouldn't be a problem, but if useful alternative cuecat software is published, there will be a huge demand for cuecats, which will make it impossible for them to be free. Actually, I think it's stupid that they are free. They should do what large catelog companies do, charge for the catelog and then refund the cost on the first order. If I buy a catelog from a company, I'm actually interested in purchasing something, and if I never purchase anything and decide to use the catelog in a place of a jackstand for working on my car, at least I know the catelog company won't try to sue me for improper use of my own property.
Basically, when you are given a "free" cuecat, you are being lied to, because you aren't allowed to use your own property. Eazel Nautilus is under the GPL license; therefore, you are allowed to use it for anything you want. You just have to give people the ability to obtain the source code, if you distribute a binary application based on the GPL'ed code. The GPL even grants you the right to charge for distribution. It just doesn't give you the right to demand licensing fees.
The idea to donate money to Eazel is based on the fact that if we benefit from it, it would be good to keep it going. Eazel isn't demanding that we use their service like DigitalConvergence is basically doing with the cuecat. The cuecat situation is as deceitful as someone giving you a car, and the next day they demand to be paid for it, not directly but by forcing you to go to their toll roads.
According to my understanding of the GPL, you only need to distribute your code modifications if you distribute the binary. If the government agency is just using this software internally, then their GPL code (and GPL infected code) will only need to be available internally, also. Of course, the question is whether the U.S. government a single entity, or would they have to distribute the source code between branches or just between federal, state, and local government agencies?
I'm basing this primarily on section 2b of the GPL.
What prevents you from releasing your code under multiple licenses? The military could use the code under a BSD license (or just plain own it), and you would also license the code under the GPL. The government would then just be restricted by not being able to incorporate the modifications by other people to your GPL'd code without it being viral to their own code.
Companies often have different licenses for their binaries and source code. A typical desktop application's license will forbid you from even trying to decompile it, but large business software may have it's source code licensed for a huge fee because their client needs extra functionality and has their own developers to implement this.
Not all American TV sci-fi is bad. Babylon 5 was great. Some episodes of Star Trek were really good, although they had a lot of fluff sometimes. The Star Trek tend to drag out themselves so long till it feels like a soap opera, although Star Trek is much more interesting when they bring back characters that died. A typical soap opera will have a separated and birth twin show up suddenly, whereas Star Trek had a temporal rift that created an alternate history to allow a daughter of Tasha Yar to be born which looked just like her. Now, THAT seems much more plausible.
Actually, in Germany you have to take classes before you can become a juror, and I think the pay is much better for jury duty.
You seem to be placing too much confidence in the logical, analytical mind of a scientist being able to make the correct decision. Scientists make mistakes all the time. The validity of scientific ideas comes through experimental verification and not through approval of geniuses. Of course, in a trial there is not time available to form hypotheses and antitheses, experiment, and repeat if necessary (shampoo science).
Also, look how many disagreements there are on/. concerning which is the best OS, programming language, text editor, etc. Being technically minded does not prevent emotional attachment to certain technologies.
Judges will never be able to be informed on all the technical issues involved in a case. It would be nice though, if there was a separate jury of scientists that could review a written record of each scientific witness' testimony and disapprove of any obvious misleading statements. That would take a lot of the effectiveness out of buying your own expert witness.
Though the Mac market is tiny compared to Windows, many more desktop applications are purchased for it than for any flavor of Unix. This might make it more attractive to develop an X application. Not only would you have the Linux/FreeBSD market, but you could get a share of the Mac market which is filled with people that are actually used to paying for software.
Of course, besides XonX, it is questionable whether there will need to be other code to glue together the Mac's print architecture with that of other Unices.
Public schools were not intended to be a daycare for kids in which to teach children respect for authority instead of their parents doing that.
The school's job is to teach. Not only is this school stifling the educational process, they are also undermining integrity and respect of authority. How eager will this child be to try to excel in school, and how much will she look up to the ideal of the pursuit of knowledge from now on?
Check out www.wxwindows.org. WxWindows is a crossplatform windowing library, which is supposed to be really cool unlike Java's AWT. It has bindings for C, Perl, Python, and probably other languages. If you use Perl or Python with WxWindows, you should have a fairly universal development platform.
I doubt that you need to access the graphics framebuffer or the audio device directly since you are actually considering using a web interface, but if you do need to create your own graphics, you can take a look at www.libsdl.org. A good example of a program using libsdl is at www.winbolo.org.
I think it would be good to discuss some of the underlying design of UNIX operating systems. In particular, the fact that almost everything on a UNIX system can be treated as a file, so you can cat out a file, you can redirect the output to another file, and the file you redirect it to can be a device such as another terminal window. You can even cat a ".au" sound file to your audio device. I think the fact that you can glue together different commands with pipes is one of the coolest and most distinguishable aspects of UNIX not found on other types of operating systems.
It is true, that programmers will be less likely to take on or complete certain projects if they don't feel they can truly capitalize on their effort, and this could reduce the number of people innovating in any particular area. However, the problem with closed source projects, is that many of the same innovations must be duplicated again and again. Companies have historically had to develop their own algorithms for a lot of the same problems because they don't have access to another companies solution. There is definitely duplication of effort in open source software, but it is often caused by a difficulty in extracting useful parts of one program for insertion into another one. A good example of the re-use of code algorithms is all the cd ripping/encoding gui utilities for Linux/*BSD. They are almost all just wrappers for some command line utilities which have already done a very good job of correcting errors when ripping a cd, and existing mp3 encoders are already well optimized, so the person who wants to write a gui interface for the job can focus on just that. The best part about open source, is that a person can switch back and forth between using whichever persons algorithm they want without worrying about wasting money by licensing yet another companies libraries. A lot of the gui interfaces for common tasks are just wrappers for command line programs, but they have the option of including the source code directly in their program to improve performance or to tailor the code for their needs without starting from scratch.
I just love the fact that the openssl libraries are no longer restricted by the RSA patent anymore. Now practically every program that communicates over TCP is adding encryption. Not only do they have free use of the algorithm since the patent has expired, but there is no need to spend a bunch of time coding and debugging their own libraries for encryption.
Though you lose some developers that might have made innovations in a given field if there was a greater economic demand for something given the absence of free alternatives, open source reduces the number of times something has to be re-innovated, and reinnovation really isn't innovation at all.
Statements that we won't switch to ebooks because they just aren't as comfortable or familiar is partially true. People that grew up on paper books are definitely going to resist switching, but if you grow up using both ebooks and paper books, you will probably be very annoyed by how many features the paper books lack. Of course, the ebooks will have to be extremely advanced so the reverse won't be true.
The last paper printed items to switch will be pulp fiction and fluff magazines, and they may never be completely supplanted. Fiction doesn't really benefit from the ebook format unless you are doing literary research, or you feel compelled to have several books with you at all time. One of the benefits of printing costing money is that it helps filter out crap, not that people don't support a lot of bad printed material no matter what, but if a magazine can afford to be printed, at least you know that it's not just the endeavor of one horrible writer with a website.
When paper books are dead, it won't be that they are wiped out, but they will be unimportant. People will still have paper books, just like some people nowadays bake bread, knit sweaters, and ride horses. In certain situations, these tasks are actually necessary, but for most of us, ebooks will be much better.
Ebooks will probably become extremely prolific after a few generations of upgrades creates a huge market of cheap used ebooks. It may still cost $400+ for a new computer system, but you can get an 80486 system for less than $100 with a monitor. Once you can get used ebooks for $10, the licensing issue will be less relavent, since you can actually lend your ebook to someone else, without just copying the data between two, which is the kind of reuse that really upsets publishing companies. Of course, the question is whether we'll actually store that much information in the ebooks then, or if everything will just be wirelessly networked, and all your data will be housed forever on a server, which means that licensing for reading a book may need to be completely different.
Your comparison of OpenSSH's name to that of JavaInvaders is completely wrong. JavaInvaders was written in Java, but it provides no functionality that makes it distinctly a Java product besides running in the Java VM. OpenSSH however provides the SSH protocol, which is it's sole purpose. The JavaInvaders name comparison would have been valid, if OpenSSH had named their product C++CommProgram or UnixCommProgram. The name OpenSSH is instrinsically related to the functionality it provides, and this makes it simple for users to know what the product claims to do.
Yahoo and Ebay are very successful companies whose names do not reflect their functionality, but they chose those names so that they could defend the name against trademark infringement, and so that their branding would be very distinguishable. If Ebay marketed themselves as Ebay Auctions, they would lose some of their distinguishablity whenever another JoeBlow Auction website went up. Yahoo now offers a lot of services beyond being an Internet directory, but it would have been very ineffective for them to market SuperDirectoryMail, etc. instead of YahooMail, etc.
Free software organizations in particular is not interested in spending a lot of money or effort marketing their software. They just want people to be able to find it, know what it's for, and try it out. There are also a lot of commercial programs out there that make sensible use of standardized names in their product names, e.g. Borland C++, McAfee VirusScan.
Think how difficult it would be to find an ssh client if they were named Bobble, George, and Sweet, and your search engine also found the utilities named Welsher, Bingan, and Sorghum that used ssh but didn't provide ssh functionality.
Microsoft is nowhere close to being dead. Hopefully it's monopolistic practices will be curbed, but the majority of the software industry revolves around them, and you are not going to be able to pick them up and move over to a new OS with any kind of simplicity. Even if Microsoft can't force computer manufacturers to only install Windows or lose special pricing or can't make their browser permanently stick to the desktop, they will still maintain the vast majority of the marketshare for desktop users.
As a server OS, Windows NT/2000 offers very few benefits, the main one is having a familiar windows interface. Since most servers will require a system administrator that understands the technical aspects beyond what can be made self-explanatory in a GUI, Linux & FreeBSD are much more attactive since not only is the OS free, but you can also use very powerfull free web servers, mail servers, SQL databases, etc. You can often install these free servers on NT, but what's the point of paying for NT then?
On the desktop, Linux still has a lot of holes to be filled, and this is going to take a considerable amount of time, and what people tend to forget is that Microsoft will have ample opportunity to improve Windows and make the change that much less compelling.
Things that Linux needs:
- Easy configuration and administration. Nautilus, Gnome, and KDE utils are nice, but they still don't all tasks as smoothly as windows.
- Better GUI performance. XFree86 4 is quite fast but I have yet to see it eat up less than 80 megs of memory, and I don't think I should need 256 MB of memory just to run netscape without swapping. Of course, as ram gets continually cheaper, the ram issue will become negligible. Maybe the GTK widgets that write directly to the video frame buffer and don't need X will be the solution.
- A no-compromise web browser. Mozilla is still a little slow, and it still chokes on some pages. It will probably be running smoothly within a year, but new plugins still surface on windows first, so this will keep it behind the curve.
- MS Office. Either run under Wine, or a competing product that doesn't sacrifice features, interoperablity, or performance. Running MS Office under Wine also keeps MS a long way from being dead, although it would tremendously reduce the need to use their OS.
- Outlook, Eudora, etc. People who write a lot of email are very attached to their email clients. Once we get acceptable options for email clients, and the points above, Linux will be very viable as a business workstation.
- Lots of Games/Tremendous game performance. There are actually people out there who are annoyed and not enchanted by the geeky accomplishment of dual-booting.
- Some huge percentage of every piece of software imaginable to compete with Windows gargantuan software market. Look how much good desktop software their is for the Mac, and still people complain about what's missing. It will probably only be possible to overcome this using Wine or some very quick emulator.
What Windows can improve on the desktop:
- Reliablity. I have heard that Win2k is a huge improvement over NT4, and if they can put this into a reasonably priced desktop version, they will be very hard to beat.
- Performance. The ram usage is too high, and there is rarely a benefit from dual procs.
- More affordable. Not likely to happen, although free OS competition might put on the pressure.
In the lists above, I am trying to show that MS has a lot less ground to cover even if they might be at a disadvantage when it comes to trying to improve their codebase without becoming incompatible with old software. The Windows software market is basically a huge mass whose inertia is going to keep it from moving unless a lot of energy is expended, and lots of energy will be wasted pushing in different directions.
As long as I can remember, I have heard people say that you should choose your software and then buy your computer. Well the OS kinda fits both into the software and the underlying platform category. You might choose your hardware for what OS it can run, but you will need to choose your OS by what software it can run. The OS (kernel + GUI) is nothing without applications.
About ten years ago on PBS's Computer Chronicles, they showed the Tony Keyboard, which was just a prototype that allowed adjustments from a normal flat keyboard up till the two halves were vertical but right next to each other.
When the show aired, the inventor was saying that he hadn't sold the idea to a manufacturer yet, because he had made the stipulation that the keyboard had to be manufactured in the United States, which no manufacturer had accepted so far.
If you are going to be using the software that you are open sourcing with the software you are selling, you should be careful with what open source license you use. If you use open source code under the GPL in your commercial applications, that software will also fall under the GPL. However, source code under the LGPL which is typically for libraries, can be used without the license infecting the rest of your source code. Only direct modifications to source code under the LGPL inherits the LGPL license.
I personally like the GPL and LGPL better than the BSD license, because it prevents people from usurping the popularity of your software, by adding features that they only available in their binary distribution, but BSD licensed source code is definitely the easiest to use without worrying about legal implications.
Slashdot Mirror
The biggest problem that I face with Postgres 7.4 is that its referential integrity's locks block INSERTS and UPDATES that should go through fine.
For example, set up these two tables:
CREATE TABLE car_type (
id serial primary key,
name varchar(20)
);
CREATE TABLE car (
id serial primary key,
car_type_id integer references car_type
);
Now, try having two different connections insert with the same foreign key value (this field does not have a unique constraint):
Connection1:
BEGIN; INSERT INTO car (car_type_id) values(1);
Connection2:
BEGIN; INSERT INTO car (car_type_id) values(1);
You will see that the second transaction is waiting for the first transaction to commit. That is just rediculous and is one reason that Postgres is still small time.
With iTunes, Final Cut Pro, and iMac, Apple obviously sees apps and services instead of the OS as the key making money off of its expertise. Even Microsoft is not secure in its OS market share and is trying to lock in users with Passport and .NET. The only downside is that the user interfaces between OSes will become even more homegenous once the OS doesn't matter.
PHP is a widely used embedded-in-html scripting language, but it is by no means the first. PHP started off as a templating language, but features were continually added which it caused it to grow beyond Server Side Includes. As such, PHP is not a well designed scripting language. If you separate your code into functions, it is extremely difficult to track down errors, because you only get the line that the error occured on and not the line that called that function.
u ct s/PerlEx/Embedded.html
PHP is definitely very easy to start programming dynamic web pages, but it is
increasingly problematic as your code base grows. If you primarily do web
design and just want a little dynamic content, PHP's shortcomings won't affect
you. However, web based applications will run into problems with PHP
repeatedly. Besides the problem where it copies objects out of an array
when you use a "foreach", you are also stuck remembering to put an amperstand
in front of every object so it gets passed or assigned by reference and not
value.
For example:
$x = $obj; # $x is a copy of $obj
$x = &$obj; # $x and $obj refer to the same object
Besides the overhead of copying large objects, it just creates one more
"gotcha" that you have to remember to avoid. It is not that good code can't
be written, but it is so much easier to write problematic code.
Other projects for embedding scripting into HTML:
http://pmz.sourceforge.net/
http://webware.sourceforge.net/
http://www.kryogenix.org/code/castalian/
http://www.zope.org/
http://aspn.activestate.com/ASPN/Reference/Prod
Basically, when you are given a "free" cuecat, you are being lied to, because you aren't allowed to use your own property. Eazel Nautilus is under the GPL license; therefore, you are allowed to use it for anything you want. You just have to give people the ability to obtain the source code, if you distribute a binary application based on the GPL'ed code. The GPL even grants you the right to charge for distribution. It just doesn't give you the right to demand licensing fees.
The idea to donate money to Eazel is based on the fact that if we benefit from it, it would be good to keep it going. Eazel isn't demanding that we use their service like DigitalConvergence is basically doing with the cuecat. The cuecat situation is as deceitful as someone giving you a car, and the next day they demand to be paid for it, not directly but by forcing you to go to their toll roads.
I can't believe they left out Weird Science. That has gotta be the most realistic hacker movie ever. When is mainstream society going to get a clue.
Starblazers was such a cool show. Why can't American cartoons have an ongoing plot lines that are longer than the two part stories on ScoobieDoo?
I'm basing this primarily on section 2b of the GPL.
What prevents you from releasing your code under multiple licenses? The military could use the code under a BSD license (or just plain own it), and you would also license the code under the GPL. The government would then just be restricted by not being able to incorporate the modifications by other people to your GPL'd code without it being viral to their own code.
Companies often have different licenses for their binaries and source code. A typical desktop application's license will forbid you from even trying to decompile it, but large business software may have it's source code licensed for a huge fee because their client needs extra functionality and has their own developers to implement this.
Not all American TV sci-fi is bad. Babylon 5 was great. Some episodes of Star Trek were really good, although they had a lot of fluff sometimes. The Star Trek tend to drag out themselves so long till it feels like a soap opera, although Star Trek is much more interesting when they bring back characters that died. A typical soap opera will have a separated and birth twin show up suddenly, whereas Star Trek had a temporal rift that created an alternate history to allow a daughter of Tasha Yar to be born which looked just like her. Now, THAT seems much more plausible.
If programming is art, is perl code abstract art? That would explain why I don't understand it when I look at it.
In a related research lab, they have been grafting adamantium onto human bones to make the person's skeleton super strong.
Actually, in Germany you have to take classes before you can become a juror, and I think the pay is much better for jury duty.
/. concerning which is the best OS, programming language, text editor, etc. Being technically minded does not prevent emotional attachment to certain technologies.
You seem to be placing too much confidence in the logical, analytical mind of a scientist being able to make the correct decision. Scientists make mistakes all the time. The validity of scientific ideas comes through experimental verification and not through approval of geniuses. Of course, in a trial there is not time available to form hypotheses and antitheses, experiment, and repeat if necessary (shampoo science).
Also, look how many disagreements there are on
Judges will never be able to be informed on all the technical issues involved in a case. It would be nice though, if there was a separate jury of scientists that could review a written record of each scientific witness' testimony and disapprove of any obvious misleading statements. That would take a lot of the effectiveness out of buying your own expert witness.
Though the Mac market is tiny compared to Windows, many more desktop applications are purchased for it than for any flavor of Unix. This might make it more attractive to develop an X application. Not only would you have the Linux/FreeBSD market, but you could get a share of the Mac market which is filled with people that are actually used to paying for software.
Of course, besides XonX, it is questionable whether there will need to be other code to glue together the Mac's print architecture with that of other Unices.
Public schools were not intended to be a daycare for kids in which to teach children respect for authority instead of their parents doing that.
The school's job is to teach. Not only is this school stifling the educational process, they are also undermining integrity and respect of authority. How eager will this child be to try to excel in school, and how much will she look up to the ideal of the pursuit of knowledge from now on?
Check out www.wxwindows.org. WxWindows is a crossplatform windowing library, which is supposed to be really cool unlike Java's AWT. It has bindings for C, Perl, Python, and probably other languages. If you use Perl or Python with WxWindows, you should have a fairly universal development platform.
I doubt that you need to access the graphics framebuffer or the audio device directly since you are actually considering using a web interface, but if you do need to create your own graphics, you can take a look at www.libsdl.org. A good example of a program using libsdl is at www.winbolo.org.
I think it would be good to discuss some of the underlying design of UNIX operating systems. In particular, the fact that almost everything on a UNIX system can be treated as a file, so you can cat out a file, you can redirect the output to another file, and the file you redirect it to can be a device such as another terminal window. You can even cat a ".au" sound file to your audio device. I think the fact that you can glue together different commands with pipes is one of the coolest and most distinguishable aspects of UNIX not found on other types of operating systems.
I think the trademark dispute of SSH in the OpenSSH name is rediculous.
Just to show how much more understanding and reasonable all the other big corporations are, I'm going to introduce the products:
OpenKleenex
OpenJello
OpenSPAM
It is true, that programmers will be less likely to take on or complete certain projects if they don't feel they can truly capitalize on their effort, and this could reduce the number of people innovating in any particular area. However, the problem with closed source projects, is that many of the same innovations must be duplicated again and again. Companies have historically had to develop their own algorithms for a lot of the same problems because they don't have access to another companies solution. There is definitely duplication of effort in open source software, but it is often caused by a difficulty in extracting useful parts of one program for insertion into another one. A good example of the re-use of code algorithms is all the cd ripping/encoding gui utilities for Linux/*BSD. They are almost all just wrappers for some command line utilities which have already done a very good job of correcting errors when ripping a cd, and existing mp3 encoders are already well optimized, so the person who wants to write a gui interface for the job can focus on just that. The best part about open source, is that a person can switch back and forth between using whichever persons algorithm they want without worrying about wasting money by licensing yet another companies libraries. A lot of the gui interfaces for common tasks are just wrappers for command line programs, but they have the option of including the source code directly in their program to improve performance or to tailor the code for their needs without starting from scratch.
I just love the fact that the openssl libraries are no longer restricted by the RSA patent anymore. Now practically every program that communicates over TCP is adding encryption. Not only do they have free use of the algorithm since the patent has expired, but there is no need to spend a bunch of time coding and debugging their own libraries for encryption.
Though you lose some developers that might have made innovations in a given field if there was a greater economic demand for something given the absence of free alternatives, open source reduces the number of times something has to be re-innovated, and reinnovation really isn't innovation at all.
Statements that we won't switch to ebooks because they just aren't as comfortable or familiar is partially true. People that grew up on paper books are definitely going to resist switching, but if you grow up using both ebooks and paper books, you will probably be very annoyed by how many features the paper books lack. Of course, the ebooks will have to be extremely advanced so the reverse won't be true.
The last paper printed items to switch will be pulp fiction and fluff magazines, and they may never be completely supplanted. Fiction doesn't really benefit from the ebook format unless you are doing literary research, or you feel compelled to have several books with you at all time. One of the benefits of printing costing money is that it helps filter out crap, not that people don't support a lot of bad printed material no matter what, but if a magazine can afford to be printed, at least you know that it's not just the endeavor of one horrible writer with a website.
When paper books are dead, it won't be that they are wiped out, but they will be unimportant. People will still have paper books, just like some people nowadays bake bread, knit sweaters, and ride horses. In certain situations, these tasks are actually necessary, but for most of us, ebooks will be much better.
Ebooks will probably become extremely prolific after a few generations of upgrades creates a huge market of cheap used ebooks. It may still cost $400+ for a new computer system, but you can get an 80486 system for less than $100 with a monitor. Once you can get used ebooks for $10, the licensing issue will be less relavent, since you can actually lend your ebook to someone else, without just copying the data between two, which is the kind of reuse that really upsets publishing companies. Of course, the question is whether we'll actually store that much information in the ebooks then, or if everything will just be wirelessly networked, and all your data will be housed forever on a server, which means that licensing for reading a book may need to be completely different.
Your comparison of OpenSSH's name to that of JavaInvaders is completely wrong. JavaInvaders was written in Java, but it provides no functionality that makes it distinctly a Java product besides running in the Java VM. OpenSSH however provides the SSH protocol, which is it's sole purpose. The JavaInvaders name comparison would have been valid, if OpenSSH had named their product C++CommProgram or UnixCommProgram. The name OpenSSH is instrinsically related to the functionality it provides, and this makes it simple for users to know what the product claims to do.
Yahoo and Ebay are very successful companies whose names do not reflect their functionality, but they chose those names so that they could defend the name against trademark infringement, and so that their branding would be very distinguishable. If Ebay marketed themselves as Ebay Auctions, they would lose some of their distinguishablity whenever another JoeBlow Auction website went up. Yahoo now offers a lot of services beyond being an Internet directory, but it would have been very ineffective for them to market SuperDirectoryMail, etc. instead of YahooMail, etc.
Free software organizations in particular is not interested in spending a lot of money or effort marketing their software. They just want people to be able to find it, know what it's for, and try it out. There are also a lot of commercial programs out there that make sensible use of standardized names in their product names, e.g. Borland C++, McAfee VirusScan.
Think how difficult it would be to find an ssh client if they were named Bobble, George, and Sweet, and your search engine also found the utilities named Welsher, Bingan, and Sorghum that used ssh but didn't provide ssh functionality.
Microsoft is nowhere close to being dead. Hopefully it's monopolistic practices will be curbed, but the majority of the software industry revolves around them, and you are not going to be able to pick them up and move over to a new OS with any kind of simplicity. Even if Microsoft can't force computer manufacturers to only install Windows or lose special pricing or can't make their browser permanently stick to the desktop, they will still maintain the vast majority of the marketshare for desktop users.
As a server OS, Windows NT/2000 offers very few benefits, the main one is having a familiar windows interface. Since most servers will require a system administrator that understands the technical aspects beyond what can be made self-explanatory in a GUI, Linux & FreeBSD are much more attactive since not only is the OS free, but you can also use very powerfull free web servers, mail servers, SQL databases, etc. You can often install these free servers on NT, but what's the point of paying for NT then?
On the desktop, Linux still has a lot of holes to be filled, and this is going to take a considerable amount of time, and what people tend to forget is that Microsoft will have ample opportunity to improve Windows and make the change that much less compelling.
Things that Linux needs:
- Easy configuration and administration. Nautilus, Gnome, and KDE utils are nice, but they still don't all tasks as smoothly as windows.
- Better GUI performance. XFree86 4 is quite fast but I have yet to see it eat up less than 80 megs of memory, and I don't think I should need 256 MB of memory just to run netscape without swapping. Of course, as ram gets continually cheaper, the ram issue will become negligible. Maybe the GTK widgets that write directly to the video frame buffer and don't need X will be the solution.
- A no-compromise web browser. Mozilla is still a little slow, and it still chokes on some pages. It will probably be running smoothly within a year, but new plugins still surface on windows first, so this will keep it behind the curve.
- MS Office. Either run under Wine, or a competing product that doesn't sacrifice features, interoperablity, or performance. Running MS Office under Wine also keeps MS a long way from being dead, although it would tremendously reduce the need to use their OS.
- Outlook, Eudora, etc. People who write a lot of email are very attached to their email clients. Once we get acceptable options for email clients, and the points above, Linux will be very viable as a business workstation.
- Lots of Games/Tremendous game performance. There are actually people out there who are annoyed and not enchanted by the geeky accomplishment of dual-booting.
- Some huge percentage of every piece of software imaginable to compete with Windows gargantuan software market. Look how much good desktop software their is for the Mac, and still people complain about what's missing. It will probably only be possible to overcome this using Wine or some very quick emulator.
What Windows can improve on the desktop:
- Reliablity. I have heard that Win2k is a huge improvement over NT4, and if they can put this into a reasonably priced desktop version, they will be very hard to beat.
- Performance. The ram usage is too high, and there is rarely a benefit from dual procs.
- More affordable. Not likely to happen, although free OS competition might put on the pressure.
In the lists above, I am trying to show that MS has a lot less ground to cover even if they might be at a disadvantage when it comes to trying to improve their codebase without becoming incompatible with old software. The Windows software market is basically a huge mass whose inertia is going to keep it from moving unless a lot of energy is expended, and lots of energy will be wasted pushing in different directions.
As long as I can remember, I have heard people say that you should choose your software and then buy your computer. Well the OS kinda fits both into the software and the underlying platform category. You might choose your hardware for what OS it can run, but you will need to choose your OS by what software it can run. The OS (kernel + GUI) is nothing without applications.
About ten years ago on PBS's Computer Chronicles, they showed the Tony Keyboard, which was just a prototype that allowed adjustments from a normal flat keyboard up till the two halves were vertical but right next to each other.
When the show aired, the inventor was saying that he hadn't sold the idea to a manufacturer yet, because he had made the stipulation that the keyboard had to be manufactured in the United States, which no manufacturer had accepted so far.
I personally like the GPL and LGPL better than the BSD license, because it prevents people from usurping the popularity of your software, by adding features that they only available in their binary distribution, but BSD licensed source code is definitely the easiest to use without worrying about legal implications.