There's always the freepascal project. I'm sure with a little hard work, it could be made to generate CLX compatible code that could be used to compile both the CLX and program itself. Since no one really knows what CLX looks like, maybe freepascal can compile this even now. Perhaps Borland could even be convinced to help with this project (since they retain copyright over the important part, CLX, anyway and can still dual license it no matter what compiler users use).
It might sound like a lot of work, but freepascal already supports many Delphi/Turbo Pascal extensions. Perhaps the Borland C++ Builder equivalent will be even better and include some kind of method of getting (some version) of GCC to generate valid code.
It's easy to badmouth a product you can't yet see or use. Just wait, maybe things won't be as bad as you think. As it stands, the only binary poorly/undocumented file that Borland's VCL products use are the.dfm file that contains the form definition. Who knows, maybe Borland will be generous enough to donate code to developers so that third party compilers (RMS license friendly) can process it just as well as Borland's official product.
Personally, I welcome Borland. They've always made quality products that worked well. My only problem with this entire thing is the free/$99 version for writing 'Open Source' programs will likely never be free enough to incorporated in some distributions such as Debian, because of the binary-only compiler (unless freepascal is/gets advanced enough to compile CLX/Kylix code). I hope that Borland does well with this endeavor. And maybe, just maybe, this'll enourage some improvements in other 'competitive' projects that are attempting to solve the same problem. I would certainly love to be able to compile CLX with GCC and FreePascal and then turn around and compile the same code in BCC and Delphi.
I hope they at least moderately change their tune about the $99 entry-level version. I think forcing people to make their personal projects 'open source' after paying for a development product is a bad idea, especially with the 'desktop' version that would allow you to release binary-only programs is marked at $999 (even the Professional versions of Borland C++ Builder and Delphi were only $499, and the $99 standard version's output was not restricted like that). The $99 pricetag should net you the dual license to CLX as well as the otherwise free compiler.
Hopefully they'll so as they've done in the past and release a $100-$150 'standard' edition that includes only the basic tools (no client/server, db and network stuff built-in). I do know they're also planning on releasing the source for certain parts of Kylix so that 'Open Source'-style apps are possible and even encouraged, whereas the IDE/RAD-tools will remain totally commercial. Not a bad model, but if they release a $100-$150 version designed for hobbyists and the general public, I'd gladly pay for it. I can personally afford $150, I can't afford $999 for a hobby I don't profit from.
Re:can Windows be beaten on the desktop?
on
Linux Is Going Down
·
· Score: 2
A $6 license would be hard to believe, unless the college/university has bought a site license cheap from Microsoft (and I would doubt they'd buy a site license for Win2K -- MS Office is a much more likely candidate). The regular educational discount for Windows 2000 Professional would be more in the area of $40-60, and for 2000 Server around $100-200. With the educational software, you're not legally bound to destroy the software when you leave campus, as you are with site licenses. You own the license for the software, although you often can't sell it outside of the academic community.
I think he's talking about design by committee in the sense that the members sit down and write the standard as a dream list of features that they'd like, ratify the standard and then go down and write the standard itself. If you want an example, look at nearly any of The Open Group's standards. DCE/RPC is my favorite. Huge standard that includes just about everything including the kitchen sink. Yes, it's powerful, but it's also very heavyweight. Every DCE server and client needs to understand nearly every possible machine combination out there. I believe it's around 250,000 lines for the reference OSF implemetation.
In this sense, there's really two different varieties of design by committee. The one that the decision makers will be the programmers, so their interests and the interests of the program coincide, and the ones that the committee represents the companies they work for, and other people will be doing the programming. In the first, you get a relatively smaller feature set that's well implemented, whereas with the latter, you usually end up with an unwieldly design that would be a pain to re-implement.
Most of those satellite systems require a telephone connection for 'Premium' services, such as PayTV and such. They could also embed signatures in certain signals that dictate which decoders are allowed to decode the signal and which aren't.
I'd be more distressed by the fact that all of this is going to do nothing more than drive costs up, since anyone who wants to build a 'licensed' display will have to pay the royalties for that decryption technology. And I imagine they probably built-in a capacity for key revokation (like DVD CSS), so should someone break into the company that made your TV set and release details on how the system works based on that manufacturer's keys, all that manufacturer's customers get to suffer. Sounds like fun, doesn't it?
I wish the MPAA would lose it's paranoia about copying and copyright infringement. It's not like a copying free-for-all has happened under the current system even without crappy MacroVision, so there's very little reason to tighten the noose. Maybe if they started charging reasonable rates for their product, people wouldn't feel the need to copy it without paying for it. I suppose the concept of 'fair use' has been a thorn in the side of organizations like the MPAA and RIAA in their neverending quest to gouge customers for every last dime, so I can see how removing that capability completely either via technology or via litigation would appeal to them.
I'm not suggesting that the man page system is better. I'm slamming Microsoft's incomplete documentation, not it's poor documentation compared to unix systems. It's frustrating looking through an API header file, seeing all these prototypes for API calls and not having any way of finding out what it does. And it's infuriating to see those same API calls used by other Microsoft products, and still be completely clueless as to what they do.
I wasn't confused by it, however my point was that I have seen many people who were. I'm not talking about the ability for a technician or an experienced users to figure it out, I'm talking about a first time user. There's no consistancy, and that's what makes a bad user interface more than anything else.
Besides, if you managed to make it to Slashdot, I doubt you're an unexperienced first-time computer user...
Personally, I think this summarizes it best. From Sun's press release:
"Microsoft has proven time and again that it is unwilling to abide by the common rules of the internet," Sueltz said. "Its behavior with regard to the Java technology was just one instance. And when presented with the choice of compatibility or termination, Microsoft chose termination."
Microsoft.NET is just another guise for Microsoft to bid on interoperability while assuring that only Microsoft official platforms are recognized as first class citizens. You can be completely sure that there will be APIs that are not published, and that other parts will be so poorly documented that they might as well not be, and so on. It will inevitably turn into Win32 over the Internet, and if you're not running Win32, you're on your own. Microsoft is only interesting in interoperability when it really equals migration to an all-Microsoft platform. The Microsoft Exchange connectors for Lotus Notes and Novell Groupwise don't exist because Microsoft has a desire to have their systems co-exist. They exist so that you can slowly convert your existing systems to Microsoft systems.
Thus, I think Corel was "dead" when the computer experiment fell through. Everything else was the icing that would have attracted people to the "cake" that was to be SideWinder. Today? They're just quibbling over the pieces...
Corel (or perhaps just former CEO Micheal Cowpland) has had a reputation with adopting technologies before they're ready, doing a rather poor job at developing the idea, and then dumping the product when it's revealed that it was unworkable anyway. Corel Wordperfect for Java is a perfect example of this. Java was still shiny and new when this was being worked on, and they even released some betas, however they failed to realize that it was going to end up both unprofitable and unworkable because of the inherent flaws in the java intpreters of the time and the language itself.
I think their Linux distribution is another example. While an admirable goal, it's still a little too soon for Linux to make any real inroads on the majority of desktops. KDE and Gnome are both very good, but there's serious questions both at the distibution/system library level and at the user-interface integration level that need to be tackled before anyone can really seriously recommend it's use to novices.
IMO, the new CEO is doing the right thing -- trimming unneeded (unprofitable) parts off the company until such time as they can afford to take such chances. Linux's shot at the end-user's desktop may yet come, but it's clear that it's not the time for that yet...
Quicktime Player had a really bad user interface, though. I really expected better from a company like Apple. It's neither intuitive or user friendly. I hate to say it, but Apple should take notes from Microsoft's Media Player 6.4 (not 7.0+), while perhaps not the prettiest is very functional and easy to understand. Microsoft's Media Player 7.0, on the other hand, falls into the same trap as Quicktime 3.0+.
There are a number of design flaws in Quicktime 3/4, not the least of which is a failure to use the native platform's widget set, instead opting for a set of bitmapped "3D-look" buttons that are mostly indistinguishable from the empty space on the UI. Another problem is the use of a 'dial' volume control, confusing first time users as how to operate it, and many believe you have to only move it a short distance in the small window that the dial appears. It also takes over several file associations without permission (Microsoft's products are no better, but this is something even Real has fixed in RealPlayer). In short, Apple tried to make it look like a physical device when it's not. They ended up wasting screen space, creating a monster UI that's completely opposite from everything they stand for that tries to take over key portions of your system.
Well said! The problem is, it's going to get worse, much worse, before it gets better. If this drag-n-drop nonsense wasn't bad enough, we've got competing object models between KDE and Gnome which means that Gimp and KWord will likely never interoperate. Gnome decided to go with CORBA, a powerful but rather heavyweight solution, whereas KDE decided to use RMI, I believe, which is much more lightweight and also lighter on features.
Maybe representatives from both sides need to start some kind of X desktop environment standards group to decide on these things, so they don't continue their spiral towards madness.
KDE and Gnome are both excellent, but it really bugs me that if I try to run a QT or KDE-based app on a Gnome desktop, it feels like it's being treated as a second class citizen, and vice versa with GTK or Gnome apps on KDE. QT and GTK both have their advantages and disadvantages when programming apps, particularily in terms of language support (GTK is more at home on C and other proceedural languages, whereas QT was designed for C++, etc)
It was this same kind of bickering back and forth that severely crippled UNIX in the past, and allowed Microsoft to make large gains in the server markets because Microsoft could dictate the standards and everyone else had to follow or be replaced by those who would.
If the teams can't come up with agreements between them about standards by themselves, maybe they should appoint a team of people to do it for them. Component and drag and drop should be the primary concern for interoperability, followed by unifying the end user experience (user interface, theming, etc) based on his or her selected desktop environment.
No, it appears he only uses OpenGL for this, which could mean there might be some very nasty slowdowns for games running on an E desktop, depending on how the glx for the card is programmed. And frankly, it doesn't really interest me. I use Linux and X on my laptop extensively, and conventionally 'anti-aliased' text just looks like blurred text on the LCD display. The X render extension on the other hand, supports subpixel rendering, which should increase the horizontal resolution (by a factor of three) that characters are rendered on that screen.
Yeah, except unlike IPv6, HDTV has been in development since the early 80s. Seeing how it took them twenty years to get it all approved for sale and broadcast, I'm figuring another 40 years before the majority of people use it.
He's saying that Kaplan doesnt like hackers? And that this colored his judgement?
No, he didn't like Martin Garbus, the defendant's lawyers. Plus, if you read the transcripts, he's continually hostile towards him and the defendants in general. He just seemed biased from the outset.
To paraphrase the introduction to the Copyright Board's ruling:
On March 19, 1998, Part VIII of the Copyright Act came into force. Until then, copying any sound recording for almost any purpose infringed copyright.
Part VIII legalizes one such activity: copying of sound recordings of musical works onto recording media for the private use of the person who makes the copy.
It does not matter whether you own the original sound recording (on any medium), you can legally make a copy for your own private use.
To emphasize this point, endnote 4 of the Copyright Board's ruling says:
Section 80 does not legalize (a) copies made for the use of someone other than the person making the copy; and (b) copies of anything else than sound recordings of musical works. It does legalize making a personal copy of a recording owned by someone else.
Well, things are a little different on a laptop computer. Almost all of the mouse pointer devices on laptops are built-in, so if you add a mouse, you do just that -- add ANOTHER mouse. If you were dealing with a desktop system, it's very different.
I would dispute that the comment was a 'troll'. He was talking about booting Linux and X on a Powerbook, which would make the lack of mouse buttons a liability rather than a feature. For X, anything less than three buttons is a problem.
Sometimes I think people are just LOOKING for reasons to be pissed off. Please try to account for the CONTEXT of the statement before ranting.
There are certain parts of X that are just plain traditional. Left click to select and copy, and middle click to paste, for instance. In some older apps, not based on either GTK or QT, such functionality was often not otherwise available. Bad UI design used to be the norm for X apps. Things have improved and still are, but you can't always escape bad designs -- even many Windows and MacOS have poorly designed interfaces. I personally believe that an GUI should be completely functional via a keyboard interface. X certainly does not meet this requirement, nor does MacOS or Windows.
Such a biting condemnation from a community that devotes space to such earth-shttering topics as hacking the Furby...
I guess you don't know how to use a modifier key? Ctl-click for contextual menus too complex a sequence for you to remember?
Well, he WAS talking about Linux and X Windows System running on a Mac, and for that purpose, one mouse button is insufficient. Even on a two button mouse and "Emulate3Buttons", X is painful to navigate. For the most part, Apple has some damn well designed machines, but for things like running X, the mouse is a limitation.
And I'm sure the button emulation for those extra two buttons is even clunkier than the "Emulate3Buttons" directive of the XF86Config for two button mouse users. Then again, as was said, it doesn't matter, with USB support you can easily buy a 'proper' X-capable mouse with anywhere from 3 to 6 buttons. The USB HID support is a wonderful thing.
The problem with this line of thinking is the fact is that it neglects the fact that the MPAA and kin do not have a literal monopoly, but they represent the interests of all the alternatives. To effectively boycott the companies behind this you'd have to stop consuming just about everything, since most of these companies have become mega-corporations with products across the board. Take a look at GE for instance, it owns NBC and a half dozen different companies making everything from consumer electronics to aircraft engines. Or Sony, that makes consumer electronics, and has both movie and music divisions. Any loss by any department in these companies would easily be compensated by the others.
If you asked the average consumer in the market for a VHS recorder what MacroVision is, you'd only get a handful that had any idea or even heard of it. Nor will most of the people in the market for DVD players know what region and CSS encoding is. The reason being is that none of the so-called safeguards against this -- journalists and the media, who's jobs it is to inform people of such issues have the authorization or space to report such stories since it's always filled with more 'glamorous' stories of crime, etc. The other issue is the same companies who are supposed to be reporting this to the average consumer are owned by the same companies that don't want this information public.
We can whine about this on Slashdot and on the EFF websites until the end of time, but unless there's a way to inform the general public via a media outlet they have access to, public outrage will never materialize and nothing will change. People are being kept in the dark on purpose, make no mistake, since consumer enlightenment of the nitty gritty details serves no purpose to those with power and influence in the media.
Well, technically, dpkg files are just ar archives of a couple.tar.gz, one containing the control (preinst, postinst, control, md5sums, etc), and data (binary).tar.gz files. At least, unlike RPMs, you can fairly easily decompose them with standard tools. Most of that document seems to talk about, what could easily be refered to as, reinventing the wheel.
He also refers to rpm and deb as first generation, which is quite false. If anything, tar.gz is first generation, ala Slackware. Not all that powerful, no dependancies at all, but it usually gets the job done. Version 1-3 of RPM could be considered second generation, as they addressed the lack of any kind of configuration in.tar.gz files and lack of dependancy information. RPMv4 and DEB could be considered third generation since they fill out the dependancy information with meaningful info, and provide some more advanced features to aid in installation.
Then again, reinventing the wheel is a UNIX tradition, and who am I to stop anyone from doing it. Energy could be spent in fixing the perceieved limitations in the other formats, but I guess there's too much bad blood between the camps to do anything of the sort.
RPM != RedHat but also RPM != rest of world, which is the thing that pisses me off with all the people with Linux blinders on.
Even for Linux users, those who do not use Red Hat or a Red Hat-derrived Linux distribution face lots of problems. Take Debian, for example. I personally feel that Debian is much cleaner than Red Hat in terms of layout, and apt-get is just plain great. On the other hand, by choosing Debian, I chose to give up RPM compatibility. To make things worse, more and more software is coming out in RPM format ONLY, especially software that's distributed in binary-format only. For example, if I want to use/try Novell's NDS products for Linux, I'm forced to use Red Hat (or compatible), because it depends on RPM so much that everything breaks under any other distro. I do have the 'alien' program for Debian, but that only goes so far, and doesn't convert pre-install and post-install scripting, as well it doesn't help when there's drastic (or not so drastic) differences in file layout (for example, RH SysV init goes in/etc/rc.d/init.d, whereas Debian uses/etc/init.d, and Slackware doesn't have either). The only benefit I can see that RPM has for binary-only distribution is that it can mostly track the installed files so they can be removed at a later date with RPM. I won't say that Debian or Slackware's package distribution method is better or worse, but rather that if your product is binary only, I'd rather see a shell script install the program than a RPM.
Slashdot Mirror
It might sound like a lot of work, but freepascal already supports many Delphi/Turbo Pascal extensions. Perhaps the Borland C++ Builder equivalent will be even better and include some kind of method of getting (some version) of GCC to generate valid code.
It's easy to badmouth a product you can't yet see or use. Just wait, maybe things won't be as bad as you think. As it stands, the only binary poorly/undocumented file that Borland's VCL products use are the .dfm file that contains the form definition. Who knows, maybe Borland will be generous enough to donate code to developers so that third party compilers (RMS license friendly) can process it just as well as Borland's official product.
Personally, I welcome Borland. They've always made quality products that worked well. My only problem with this entire thing is the free/$99 version for writing 'Open Source' programs will likely never be free enough to incorporated in some distributions such as Debian, because of the binary-only compiler (unless freepascal is/gets advanced enough to compile CLX/Kylix code). I hope that Borland does well with this endeavor. And maybe, just maybe, this'll enourage some improvements in other 'competitive' projects that are attempting to solve the same problem. I would certainly love to be able to compile CLX with GCC and FreePascal and then turn around and compile the same code in BCC and Delphi.
I hope they at least moderately change their tune about the $99 entry-level version. I think forcing people to make their personal projects 'open source' after paying for a development product is a bad idea, especially with the 'desktop' version that would allow you to release binary-only programs is marked at $999 (even the Professional versions of Borland C++ Builder and Delphi were only $499, and the $99 standard version's output was not restricted like that). The $99 pricetag should net you the dual license to CLX as well as the otherwise free compiler.
Hopefully they'll so as they've done in the past and release a $100-$150 'standard' edition that includes only the basic tools (no client/server, db and network stuff built-in). I do know they're also planning on releasing the source for certain parts of Kylix so that 'Open Source'-style apps are possible and even encouraged, whereas the IDE/RAD-tools will remain totally commercial. Not a bad model, but if they release a $100-$150 version designed for hobbyists and the general public, I'd gladly pay for it. I can personally afford $150, I can't afford $999 for a hobby I don't profit from.
A $6 license would be hard to believe, unless the college/university has bought a site license cheap from Microsoft (and I would doubt they'd buy a site license for Win2K -- MS Office is a much more likely candidate). The regular educational discount for Windows 2000 Professional would be more in the area of $40-60, and for 2000 Server around $100-200. With the educational software, you're not legally bound to destroy the software when you leave campus, as you are with site licenses. You own the license for the software, although you often can't sell it outside of the academic community.
In this sense, there's really two different varieties of design by committee. The one that the decision makers will be the programmers, so their interests and the interests of the program coincide, and the ones that the committee represents the companies they work for, and other people will be doing the programming. In the first, you get a relatively smaller feature set that's well implemented, whereas with the latter, you usually end up with an unwieldly design that would be a pain to re-implement.
I'd be more distressed by the fact that all of this is going to do nothing more than drive costs up, since anyone who wants to build a 'licensed' display will have to pay the royalties for that decryption technology. And I imagine they probably built-in a capacity for key revokation (like DVD CSS), so should someone break into the company that made your TV set and release details on how the system works based on that manufacturer's keys, all that manufacturer's customers get to suffer. Sounds like fun, doesn't it?
I wish the MPAA would lose it's paranoia about copying and copyright infringement. It's not like a copying free-for-all has happened under the current system even without crappy MacroVision, so there's very little reason to tighten the noose. Maybe if they started charging reasonable rates for their product, people wouldn't feel the need to copy it without paying for it. I suppose the concept of 'fair use' has been a thorn in the side of organizations like the MPAA and RIAA in their neverending quest to gouge customers for every last dime, so I can see how removing that capability completely either via technology or via litigation would appeal to them.
I'm not suggesting that the man page system is better. I'm slamming Microsoft's incomplete documentation, not it's poor documentation compared to unix systems. It's frustrating looking through an API header file, seeing all these prototypes for API calls and not having any way of finding out what it does. And it's infuriating to see those same API calls used by other Microsoft products, and still be completely clueless as to what they do.
Besides, if you managed to make it to Slashdot, I doubt you're an unexperienced first-time computer user...
I think their Linux distribution is another example. While an admirable goal, it's still a little too soon for Linux to make any real inroads on the majority of desktops. KDE and Gnome are both very good, but there's serious questions both at the distibution/system library level and at the user-interface integration level that need to be tackled before anyone can really seriously recommend it's use to novices.
IMO, the new CEO is doing the right thing -- trimming unneeded (unprofitable) parts off the company until such time as they can afford to take such chances. Linux's shot at the end-user's desktop may yet come, but it's clear that it's not the time for that yet...
There are a number of design flaws in Quicktime 3/4, not the least of which is a failure to use the native platform's widget set, instead opting for a set of bitmapped "3D-look" buttons that are mostly indistinguishable from the empty space on the UI. Another problem is the use of a 'dial' volume control, confusing first time users as how to operate it, and many believe you have to only move it a short distance in the small window that the dial appears. It also takes over several file associations without permission (Microsoft's products are no better, but this is something even Real has fixed in RealPlayer). In short, Apple tried to make it look like a physical device when it's not. They ended up wasting screen space, creating a monster UI that's completely opposite from everything they stand for that tries to take over key portions of your system.
Maybe representatives from both sides need to start some kind of X desktop environment standards group to decide on these things, so they don't continue their spiral towards madness.
KDE and Gnome are both excellent, but it really bugs me that if I try to run a QT or KDE-based app on a Gnome desktop, it feels like it's being treated as a second class citizen, and vice versa with GTK or Gnome apps on KDE. QT and GTK both have their advantages and disadvantages when programming apps, particularily in terms of language support (GTK is more at home on C and other proceedural languages, whereas QT was designed for C++, etc)
It was this same kind of bickering back and forth that severely crippled UNIX in the past, and allowed Microsoft to make large gains in the server markets because Microsoft could dictate the standards and everyone else had to follow or be replaced by those who would.
If the teams can't come up with agreements between them about standards by themselves, maybe they should appoint a team of people to do it for them. Component and drag and drop should be the primary concern for interoperability, followed by unifying the end user experience (user interface, theming, etc) based on his or her selected desktop environment.
No, it appears he only uses OpenGL for this, which could mean there might be some very nasty slowdowns for games running on an E desktop, depending on how the glx for the card is programmed. And frankly, it doesn't really interest me. I use Linux and X on my laptop extensively, and conventionally 'anti-aliased' text just looks like blurred text on the LCD display. The X render extension on the other hand, supports subpixel rendering, which should increase the horizontal resolution (by a factor of three) that characters are rendered on that screen.
Yeah, except unlike IPv6, HDTV has been in development since the early 80s. Seeing how it took them twenty years to get it all approved for sale and broadcast, I'm figuring another 40 years before the majority of people use it.
Quote from that page:
To paraphrase the introduction to the Copyright Board's ruling:
It does not matter whether you own the original sound recording (on any medium), you can legally make a copy for your own private use.To emphasize this point, endnote 4 of the Copyright Board's ruling says:
Well, things are a little different on a laptop computer. Almost all of the mouse pointer devices on laptops are built-in, so if you add a mouse, you do just that -- add ANOTHER mouse. If you were dealing with a desktop system, it's very different.
Sometimes I think people are just LOOKING for reasons to be pissed off. Please try to account for the CONTEXT of the statement before ranting.
There are certain parts of X that are just plain traditional. Left click to select and copy, and middle click to paste, for instance. In some older apps, not based on either GTK or QT, such functionality was often not otherwise available. Bad UI design used to be the norm for X apps. Things have improved and still are, but you can't always escape bad designs -- even many Windows and MacOS have poorly designed interfaces. I personally believe that an GUI should be completely functional via a keyboard interface. X certainly does not meet this requirement, nor does MacOS or Windows.
And I'm sure the button emulation for those extra two buttons is even clunkier than the "Emulate3Buttons" directive of the XF86Config for two button mouse users. Then again, as was said, it doesn't matter, with USB support you can easily buy a 'proper' X-capable mouse with anywhere from 3 to 6 buttons. The USB HID support is a wonderful thing.
If you asked the average consumer in the market for a VHS recorder what MacroVision is, you'd only get a handful that had any idea or even heard of it. Nor will most of the people in the market for DVD players know what region and CSS encoding is. The reason being is that none of the so-called safeguards against this -- journalists and the media, who's jobs it is to inform people of such issues have the authorization or space to report such stories since it's always filled with more 'glamorous' stories of crime, etc. The other issue is the same companies who are supposed to be reporting this to the average consumer are owned by the same companies that don't want this information public.
We can whine about this on Slashdot and on the EFF websites until the end of time, but unless there's a way to inform the general public via a media outlet they have access to, public outrage will never materialize and nothing will change. People are being kept in the dark on purpose, make no mistake, since consumer enlightenment of the nitty gritty details serves no purpose to those with power and influence in the media.
He also refers to rpm and deb as first generation, which is quite false. If anything, tar.gz is first generation, ala Slackware. Not all that powerful, no dependancies at all, but it usually gets the job done. Version 1-3 of RPM could be considered second generation, as they addressed the lack of any kind of configuration in .tar.gz files and lack of dependancy information. RPMv4 and DEB could be considered third generation since they fill out the dependancy information with meaningful info, and provide some more advanced features to aid in installation.
Then again, reinventing the wheel is a UNIX tradition, and who am I to stop anyone from doing it. Energy could be spent in fixing the perceieved limitations in the other formats, but I guess there's too much bad blood between the camps to do anything of the sort.
Even for Linux users, those who do not use Red Hat or a Red Hat-derrived Linux distribution face lots of problems. Take Debian, for example. I personally feel that Debian is much cleaner than Red Hat in terms of layout, and apt-get is just plain great. On the other hand, by choosing Debian, I chose to give up RPM compatibility. To make things worse, more and more software is coming out in RPM format ONLY, especially software that's distributed in binary-format only. For example, if I want to use/try Novell's NDS products for Linux, I'm forced to use Red Hat (or compatible), because it depends on RPM so much that everything breaks under any other distro. I do have the 'alien' program for Debian, but that only goes so far, and doesn't convert pre-install and post-install scripting, as well it doesn't help when there's drastic (or not so drastic) differences in file layout (for example, RH SysV init goes in /etc/rc.d/init.d, whereas Debian uses /etc/init.d, and Slackware doesn't have either). The only benefit I can see that RPM has for binary-only distribution is that it can mostly track the installed files so they can be removed at a later date with RPM. I won't say that Debian or Slackware's package distribution method is better or worse, but rather that if your product is binary only, I'd rather see a shell script install the program than a RPM.