You're dense. The technologies may live on, but the standards process for these technologies is dead -- which is the topic for this thread. Even Posix, certainly the most important of the given standards, is hardly a lively and active standard.
Nice strawman -- the standard of someone who is good at arguing, but bad at reading.
No doubt like many of us, I've had RSI problems, but my condition has improved dramatically over the last couple years (without decreasing my computer use).
For a while I thought my keyboard (Kinesis Ergo) was to thank (and perhaps the MS Trackball Explorer, which you click with your thumb). I still like my keyboard -- RSI aside, it's a great input device -- but I think there's a limit to what they can do.
What really helped me was the armless chair I got, and the posture that has encouraged me to take. Which is actually no posture at all -- I shift positions on the chair at least twice an hour, sometimes leaning sideways, sometimes forward, and reclined in different ways. There's no real stability to the chair. None of the postures I take are ergonomically correct postures, which is why I think it's good -- no posture is right for too long a period.
The problem with a fancy system like this is that it's all about the Right Posture. It creates a whole frame around some "perfect" position, and from the look of it you'd have a hard time taking any other position. It's the same with a lot of the ergonomic devices, which advertise the relaxed and supported position you take, but you are locked into a single position, so even if there's less damage you have to worry that eventually it will accumulate since it's always the same damage.
First, to clarify: standards help mitigate the problems with monopolists, they do not help monopolists.
Documentation is certainly important at a certain stage. The useful standards you do note are all fairly informal, embodied in RFCs, a process that encourages (demands?) implementation before specification, and the process does not look anything the formalism of, say, Posix.
POSIX, NFS, DCE, CDE/Motif, X11, Kerberos, etc etc etc. How can you not have heard of these?
Yes, and since proprietary Unices have faded away all these standards have become stagnant or even dead (e.g., CDE). That was my point.
It amazes me that Slashbots can criticize players like Microsoft for ignoring standards when it suits them, then turn around and do exactly the same thing themselves.
Open Source code is its own standard. Standards are for secretive companies, for companies that don't trust each other, and for monopolists. Standards are a poor surrogate for an open implementation, but in the proprietary world that's all you get. But we don't need them, the implementation is the standard, and it's almost always a more complete and completely specified standard than any paper standard.
There hasn't been much movement on formal standards, at least among Unices. FHS is the closest thing I can think of recently. I think that's because the only Unices that have any significant forward movement are Open Source/Free Software. There are de facto standards being created, some intentionally, some may only be recognizable as standards after the fact. But there's little need for formality, trademarks, committees, or certifications.
I'm having a hard time figuring out how that would work. What's the value of the stock, or the company, if you can't sell the stock? I guess you can, amongst the other employees, but that's a limited market, and creates a circular system -- if SAIC ever did get in trouble, stock prices would be hit doubly hard, since stock would be worth less, and the potential market would be making less.
Whatever it is, I don't think it's really "stock", it's some sort of limited collective ownership and profit sharing.
You are certainly right, that SDI is a political, not physical tool. At least as it is proposed.
It seems very likely to me that SDI is just a cover for the militarization of space. Current treaties and general agreements keep the US from putting offensive weapons into space. SDI, however, would be much more effective as an offensive weapon than a defensive one -- attacking unprepared targets from space is far easier than prepared targets (like a missile), and the kind of weapons used to take down a missile should be able to kill a person from space as well.
Politically this makes more sense to me -- other countries are very much opposed to this not because they worry about the US becoming inpervious to missiles (they know it is not a practical goal, and they don't care to attack the US anyway). An offensive weapon from space, however, should be extremely troubling to anyone.
Military research is a waste. What good does classified research do for us? Pay for intelligent people to work on problems, the solutions to which we'll only see when it is deamed irrelevent... that's not a smart deal.
The point is you can never test SDI, because you are working against an opponent that is consciously trying to work around your system. You can never predict how the attack with occur. Then you can never simulate the attack, even as you might predict it -- you can never launch empty missiles at a realistic target. Instead at best you do tests over the ocean. That's why it will always be in beta, which is not a useful status for a safeguard.
But more concerning is the fact that despite their effort they cannot pass even their minimal tests, and resort to fraud instead. We have tried, and failed. The whole thing is military graft -- money being sent down a pit to profit defense companies. They probably hope to cover up the failure of the system by avoiding any real-world test of the system, though certainly avoiding having missiles launched at the US is a good goal regardless.
Apple uses Software Update to upgrade kernels, daemons (Apache, Samba, BIND, and Sendmail are recent examples), libraries (OpenSSL, several times). It's not just used to upgrade a few Apple applications. I have no reason to believe they cannot update the entire OS (except third-party apps) with it. Do you?
The granularity of Software Update is extremely crude. It doesn't update BIND or Sendmail, it has an "operating system" update. That's nothing like what Debian provides. It's just a big fat updater like Microsoft uses. The result is probably something on the order of 1/100th of the number of discrete packages as Debian.
I'm not saying apt-get has a great interface. Specifically, the repository needs to have more editorial annotations -- that's probably one of the big things Lindows adds. That, combined with an interface that knew a little more about what to hide (e.g, the installation of lib* packages), would be a nice system. That does not change the fact that apt-get/dpkg provides a foundation that is far more powerful than Software Update, and yes, to the point that Software Update's interface would fall apart when trying to present everything that apt-get provides.
Are you familiar with the MacOS X "application bundle"? Bundles can be copied anywhere in the system, so there's no question of "installed in the proper location". They are self-contained, so they don't affect other software. Cleanly remove? Drag it to the trash can.
That works okay for large, monolithic, proprietary applications. File paths are a way (on Linux) that software finds out about the system, about itself, and about other software (on Windows it would be the registry -- which is just as static as the Linux file system, only it's flawed in that it's not managed). OS X applications that want to become part of a larger system must also hook in at some point -- either in specific places in the filesystem (probably the Unixy part of the filesystem), or as in Windows using registry-like facilities (netinfo, I think).
Debian packages work together beautifully. Only with tight organizational cooperation does this happen in the proprietary world, like among the Apple iLife applications, or MS Office. Arranging this requires introspection and standards that bundles cannot provide.
An interesting compromise (between Lindows and OS X) would be if a user at the console (with administrator priviledge) could permit certain (all?) operations just with a normal confirmation (click OK) -- not entering any password.
Supposing the dialog was implemented so it was somewhat secure (i.e., a program couldn't fake you clicking OK), and distinct (so that it was a signal you were doing something that required trust in the program), and not used excessively (so it didn't become second nature to click the OK)... it seems like it'd be pretty decent security for a personal machine.
the first person to touch it can do whatever they want.
In contrast, however, this is true for MacOS X, and for any version of Linux (because the person that sets up the machine gets to set up the root account).
But, in contrast to Linux, MacOS X doesn't require root priviledges to change the clock or a number of other operations (many hardware related). Linux often requires higher permissions than it should, because it is coming from the perspective of a shared, multiuser environment, or it does not give the console user extra priviledges.
Software Update is easy, but it's not comparable to apt-get or the Debian package repository.
Software Update -- much like Windows Update -- handles specific software on the computer, and as far as I can tell just Apple software. It is not anywhere close to inclusive, and it seems to have no concept of the computer as a system.
apt-get, OTOH, knows every detail of the software on my system. It upgrades everything, not just the software of a select vendor, or at a certain level of the operating system. It protects me from incompatibilities across all my software.
Sure, Software Update is easier, but it's nowhere near as powerful as apt-get. And it's only slightly easier. It has a GUI. Apt-get has GUIs, though I wouldn't claim they were easier -- but that's because Software Update deals with such a small set of software that it can present the most simplistic interface and get away with it. If every piece of software on your computer was handled by Software Update it wouldn't be so easy anymore.
Apt-get is awesome. I can go to some strange Debian box that I know little about, and (presuming I have sudo or root access) I can safely install software, even trivially small software, without worry -- I know that it will be installed in the proper location, that it won't affect other software on the system, and that someone later can safely and cleanly remove it if they choose. MacOS X can't compare. (Well, maybe with fink:)
I guess you can draw a line between that kind of egalitarian thought and the dictionary definition of communism, but its kind of meaningless, since all the communist states I know of are from the start run by a clique which took control and ruthlessly maintained it pretty much indistinguishably from a Fascist state.
It's because you just have Soviet (and Soviet-influenced) and Maoist states to think of. Two examples aren't enough to go on. Communist power structures are not uncommon in history, or even in the present, but they generally happen on a much smaller scale.
The core communist concept: from each by their ability, to each by their need.
The communist structure is very common in marriages, nuclear families, extended families, small communities... the less people, the more connected they are, the more this communist principle is carried out. Small to medium sized communist communities are quite common in history.
Some of the same problems that communist communities have scaling are probably the same problems Free Software projects have -- though of course without the economics of scarcity, Free Software goes much further before it hits those barriers.
Re:weeds aren't the problem, weed killer is.
on
Hi-Tech Weed-Killer
·
· Score: 1
Plant cover crops in-between and among your primary crop. It could be a harvestable plant, such as pole beans on corn or basil with tomatoes, though this makes harvesting a job for people and not machines.
But SCO isn't producing technology of any worth. They have no incentive to play nice -- they are a dying company (or already dead?). Their operating system software is worthless, which is pretty much what they are claiming in the lawsuit -- though they say it is worthless because of theft, not superior competition.
While open formats are a certainly a good idea, Open Source/Free Software implies far more than an open format, and is a mandate with social, not just functional implications.
I argue elsewhere that open formats can only be truly open with viewable source, though a well-maintained reference implementation may be sufficient. Standards without an accurate reference implementation are fake, IMHO.
But I would go further -- investing in Open Source is a long-term investment in the public good. The principles of Free Software are principles of the public good, and that's the business the government is in.
But in order for government to take the value of the public good into account, the decision has to be made on a political level, not by the individual departments or IT staff. Departments and other governmental organizations are typically organized to specialize on one thing, or one aspect of the public good. This encourages expedient solutions which may not take into account larger issues.
OpenOffice's documentation, however, will always be available from openoffice.org.
More importantly, a completely accurate description of the semantics of OpenOffice documents is available through the OO source code. This is in contrast to Microsoft's document specifications, which are not guaranteed to be correct -- they are often incorrect simply because specifications are just as capable of being buggy as programs are (but there's no incentive for MS to debug the specification).
This is a fundamental openess that is only available when the source is freely available. It doesn't necessarily have to be true Open Source to have this benefit, i.e., freely redistributable, but you must be able to read the code and make personal modifications without being under an NDA.
I think this is very important -- it is clear that Microsoft and other vendors (e.g., Sun) do not create or support open specifications in good will. For example, NT's support of the Posix standard (among many such examples).
(Specifically Microsoft, being a criminal and perjurous corporation, must be viewed with suspicion in all it says and does. Trusting Microsoft with long-term strategies is like trusting a convict to collect credit card information, or trusting John Poindexter with our privacy... which is to say stupid but typical)
To all the other repliers who are saying "But PostgreSQL/MySQL doesn't have feature X" or "PostgreSQL/MySQL is still just playing catch-up", I think you're missing the point.
Yes, they are playing catch-up (MySQL even moreso). But they long ago surpassed Oracle in other areas. While I don't say this from experience, it seems clear that administration and management of either of the free databases is far easier than Oracle. They also scale in a way that Oracle does not -- they scale down.
The OSS databases aren't going to beat Oracle at the reliability, enterprise, etc. They may be able to be good enough, which is what's important. They already have beaten Oracle at maintainability.
The same is true of Linux as well. Linux isn't more reliable or more powerful than the proprietary Unices. But it's far more manageable, easier to administer, and it scales down better than those other operating systems. Those other systems are dying... sure, they'll be around for a long time, but they've become a legacy. No doubt Oracle will be around for a very long time, but if PostgreSQL or MySQL can become good enough in Oracle's arena, then Oracle runs a serious risk of becoming a legacy database.
It seems like they are targeting server-side templating as well, with common navigational elements surrounding the (dynamic?) content.
Pprior art for this should be even easier to find. NCSA httpd had server-side includes, which are primarily used for this purpose. But maybe they are specifically thinking of dynamic content, in which case specific CGI software, or software that created static HTML pages, would be the proper prior art.
For example, magazine ads are relative benign, you don't see people rising up demanding ways to get around magazine ads... Those ads are intrusive, in that they're always there in front of the reader, but they're not too annoying.
I find them quite annoying. I can't easily leaf through a magazine, because different weight papers are used to divert my finger to certain pages (never the ones I want). I can't find the contents because it's hidden behind some random number of ads in the front of the magazine. And once I do find the contents, I can't find the article because only about half the pages have numbers on them (since ads don't have numbers) -- worse when the magazine decides that some ad section is special, and isn't included in the page count, so there's fifteen pages between "page 94" and "page 95".
So there, I can bitch about all ads, all the time if I want to! I can't do as much about the magazine ads, though...
Really, though, let's not pretend that ads in our real life aren't without their cost.
Linus has made an exception for binary kernel modules, though that's not popular with all people. His opinion is based on the idea that it would be excessive if anyone who used a system call would have to obey the GPL, including kernel modules.
This is not the standard interpretation, and does not apply to other GPLed software.
You are not allowed to use LGPL as a static library, without falling under the GPL. Some LGPLed software has specific exceptions to this (for instance, in places where it's expected to be used in an embedded environment). I believe, for instance, that Guile has this exception in its license that allows you to use it as a static library without falling under the GPL.
That's just silly. For instance, on my Debian machine I can install just mozilla-browser, or optionally the features of mailnews, xmlterm, or chatzilla. If I choose to install mailnews after installing mozilla-browser, I don't have to reinstall my browser. Pieces can be upgraded incrementally.
That's an easy situation. What about PHP? To make PHP install nicely, it has to know I'm using Apache, and how Apache is configured, and it has to make Apache know that.php means PHP should handle the file. On Debian it does this, all on its own. To do this without a package manager (and good policies), the OS either has to have strong policy (which means you can't just drag the application anywhere and use it), or it can have a registry, which is just a way of hiding the crud, and we all know how bad an idea that is. Or you're stuck. If PHP can't detect where Apache lives, you're stuck, your installation process will suck.
On a system with rich functionality, extensible software, and widely shared libraries and modules, drag-and-drop installation is a step backward. I have 1500 packages installed on my computer, that's only possible because of a great package management system, with developers backing it up.
I think people who want drap-and-drop installs haven't used Debian. Installing Debian packages is easy and highly reliable, even as it handles the entire system cleanly and without undue redundancy.
Sure, Debian packages could stand to be better annotated. The average user doesn't care about all the libraries that are installed, and a good installer would hide these packages (for installation, selection, and upgrading). I assume Lindows does something like this with its service. The essential system underneith Debian is far more advanced than anything Windows or Mac has.
When I first read about MacOS X's take on the Unix filesystem, I thought Great! Finally something that makes a bit more sense.
But after using it, I think it's mostly time wasted. The system's filesystem doesn't need to look pretty -- rather, the user shouldn't see it at all. When you go up directories as far as you can, you should not reach some overriding root directory in which everything is stored.
There's two very different things that a filesystem is used for -- user files, and application/system files. There's no reason for these to use the same metaphors or to be accessible through the same mechanisms.
Application files are usually useless out of context. They belong to the application, are updated at its discretion, and all operations on them are done by the application (perhaps at the user's request). We keep them in a hierarchical filesystem, keyed by strings separated with slashes, because that works well enough and it's quite fast (compared to an RDBMS, which would be silly, since its features would be wasted). "Applications" should be read to include the essential system software as well as end-user applications.
User files are something else entirely. They have nothing to do with application files, except that they are traditionally implemented the same way. User files aren't updated or controlled by the system, they aren't sorted by the system, they don't belong to the system or to an application (though they are manipulated through applications). They warrant the overhead of arbitrary metadata, versioning, and regular backups. RDBMS concepts could be usefully applied to these files.
Object oriented concept should also be applied, as opposed to the everything-is-a-stream-of-bytes metaphor that the traditional filesystem uses. This provides a basis upon which more complex data can be presented through the filesystem metaphor -- turning an mbox file into a folder, presenting pieces of hardware as files. There's no reason to bother actually presenting these as traditional files (like/hardware), because they are only usable with intermediary programs. Let those programs present these as files in this higher-level user filesystem.
The conclusion I gain from this is that this work doesn't need to be done on the filesystem level. Let package management systems deal with that -- they do so effectively. Instead the applications that a user interacts with need to present this totally new filesystem, which incidentally stores its information on the old filesystem (though every user file may actually be implemented as a directory, for instance). Gnome VFS is along these lines, and I believe KDE offers a similar abstraction, though neither is as ambitious as what I described... but maybe they'll continue to move in that direction. Nautilus had some of these ambitions as well, though I always felt it fell far short, even as it was bogged down by the features it did have.
Anyway, the New OS should happen in the desktop application arena. Traditional Unix (like Linux as implemented by Debian or Redhat) works well and is a good foundation. People just need the ambition to build new metaphors ontop of that foundation, not try to port the foundation to the new metaphor itself.
I've long wanted someone to use create something like a Mac (hardware-wise), only with Linux -- hardware that works really well together, and works well with Linux (preinstalled), and is reliable: a computer that Just Works.
When Lindows started being bundled with cheap computers I was optimistic that this might happen -- but the reviews seem to agree that the hardware is underpowered and not very good quality.
Do you have any plans on offering higher-end computers (particularly higher quality), branded specifically for Lindows? Plans for branded peripherals that work seemlessly with that hardware? Computers that are aesthetically pleasing?
Slashdot Mirror
Nice strawman -- the standard of someone who is good at arguing, but bad at reading.
For a while I thought my keyboard (Kinesis Ergo) was to thank (and perhaps the MS Trackball Explorer, which you click with your thumb). I still like my keyboard -- RSI aside, it's a great input device -- but I think there's a limit to what they can do.
What really helped me was the armless chair I got, and the posture that has encouraged me to take. Which is actually no posture at all -- I shift positions on the chair at least twice an hour, sometimes leaning sideways, sometimes forward, and reclined in different ways. There's no real stability to the chair. None of the postures I take are ergonomically correct postures, which is why I think it's good -- no posture is right for too long a period.
The problem with a fancy system like this is that it's all about the Right Posture. It creates a whole frame around some "perfect" position, and from the look of it you'd have a hard time taking any other position. It's the same with a lot of the ergonomic devices, which advertise the relaxed and supported position you take, but you are locked into a single position, so even if there's less damage you have to worry that eventually it will accumulate since it's always the same damage.
Documentation is certainly important at a certain stage. The useful standards you do note are all fairly informal, embodied in RFCs, a process that encourages (demands?) implementation before specification, and the process does not look anything the formalism of, say, Posix.
Yes, and since proprietary Unices have faded away all these standards have become stagnant or even dead (e.g., CDE). That was my point.There hasn't been much movement on formal standards, at least among Unices. FHS is the closest thing I can think of recently. I think that's because the only Unices that have any significant forward movement are Open Source/Free Software. There are de facto standards being created, some intentionally, some may only be recognizable as standards after the fact. But there's little need for formality, trademarks, committees, or certifications.
Whatever it is, I don't think it's really "stock", it's some sort of limited collective ownership and profit sharing.
It seems very likely to me that SDI is just a cover for the militarization of space. Current treaties and general agreements keep the US from putting offensive weapons into space. SDI, however, would be much more effective as an offensive weapon than a defensive one -- attacking unprepared targets from space is far easier than prepared targets (like a missile), and the kind of weapons used to take down a missile should be able to kill a person from space as well.
Politically this makes more sense to me -- other countries are very much opposed to this not because they worry about the US becoming inpervious to missiles (they know it is not a practical goal, and they don't care to attack the US anyway). An offensive weapon from space, however, should be extremely troubling to anyone.
Military research is a waste. What good does classified research do for us? Pay for intelligent people to work on problems, the solutions to which we'll only see when it is deamed irrelevent... that's not a smart deal.
But more concerning is the fact that despite their effort they cannot pass even their minimal tests, and resort to fraud instead. We have tried, and failed. The whole thing is military graft -- money being sent down a pit to profit defense companies. They probably hope to cover up the failure of the system by avoiding any real-world test of the system, though certainly avoiding having missiles launched at the US is a good goal regardless.
I'm not saying apt-get has a great interface. Specifically, the repository needs to have more editorial annotations -- that's probably one of the big things Lindows adds. That, combined with an interface that knew a little more about what to hide (e.g, the installation of lib* packages), would be a nice system. That does not change the fact that apt-get/dpkg provides a foundation that is far more powerful than Software Update, and yes, to the point that Software Update's interface would fall apart when trying to present everything that apt-get provides.
That works okay for large, monolithic, proprietary applications. File paths are a way (on Linux) that software finds out about the system, about itself, and about other software (on Windows it would be the registry -- which is just as static as the Linux file system, only it's flawed in that it's not managed). OS X applications that want to become part of a larger system must also hook in at some point -- either in specific places in the filesystem (probably the Unixy part of the filesystem), or as in Windows using registry-like facilities (netinfo, I think).Debian packages work together beautifully. Only with tight organizational cooperation does this happen in the proprietary world, like among the Apple iLife applications, or MS Office. Arranging this requires introspection and standards that bundles cannot provide.
Supposing the dialog was implemented so it was somewhat secure (i.e., a program couldn't fake you clicking OK), and distinct (so that it was a signal you were doing something that required trust in the program), and not used excessively (so it didn't become second nature to click the OK)... it seems like it'd be pretty decent security for a personal machine.
But, in contrast to Linux, MacOS X doesn't require root priviledges to change the clock or a number of other operations (many hardware related). Linux often requires higher permissions than it should, because it is coming from the perspective of a shared, multiuser environment, or it does not give the console user extra priviledges.
Software Update -- much like Windows Update -- handles specific software on the computer, and as far as I can tell just Apple software. It is not anywhere close to inclusive, and it seems to have no concept of the computer as a system.
apt-get, OTOH, knows every detail of the software on my system. It upgrades everything, not just the software of a select vendor, or at a certain level of the operating system. It protects me from incompatibilities across all my software.
Sure, Software Update is easier, but it's nowhere near as powerful as apt-get. And it's only slightly easier. It has a GUI. Apt-get has GUIs, though I wouldn't claim they were easier -- but that's because Software Update deals with such a small set of software that it can present the most simplistic interface and get away with it. If every piece of software on your computer was handled by Software Update it wouldn't be so easy anymore.
Apt-get is awesome. I can go to some strange Debian box that I know little about, and (presuming I have sudo or root access) I can safely install software, even trivially small software, without worry -- I know that it will be installed in the proper location, that it won't affect other software on the system, and that someone later can safely and cleanly remove it if they choose. MacOS X can't compare. (Well, maybe with fink :)
The core communist concept: from each by their ability, to each by their need.
The communist structure is very common in marriages, nuclear families, extended families, small communities... the less people, the more connected they are, the more this communist principle is carried out. Small to medium sized communist communities are quite common in history.
Some of the same problems that communist communities have scaling are probably the same problems Free Software projects have -- though of course without the economics of scarcity, Free Software goes much further before it hits those barriers.
But SCO isn't producing technology of any worth. They have no incentive to play nice -- they are a dying company (or already dead?). Their operating system software is worthless, which is pretty much what they are claiming in the lawsuit -- though they say it is worthless because of theft, not superior competition.
I argue elsewhere that open formats can only be truly open with viewable source, though a well-maintained reference implementation may be sufficient. Standards without an accurate reference implementation are fake, IMHO.
But I would go further -- investing in Open Source is a long-term investment in the public good. The principles of Free Software are principles of the public good, and that's the business the government is in.
But in order for government to take the value of the public good into account, the decision has to be made on a political level, not by the individual departments or IT staff. Departments and other governmental organizations are typically organized to specialize on one thing, or one aspect of the public good. This encourages expedient solutions which may not take into account larger issues.
This is a fundamental openess that is only available when the source is freely available. It doesn't necessarily have to be true Open Source to have this benefit, i.e., freely redistributable, but you must be able to read the code and make personal modifications without being under an NDA.
I think this is very important -- it is clear that Microsoft and other vendors (e.g., Sun) do not create or support open specifications in good will. For example, NT's support of the Posix standard (among many such examples).
(Specifically Microsoft, being a criminal and perjurous corporation, must be viewed with suspicion in all it says and does. Trusting Microsoft with long-term strategies is like trusting a convict to collect credit card information, or trusting John Poindexter with our privacy... which is to say stupid but typical)
Yes, they are playing catch-up (MySQL even moreso). But they long ago surpassed Oracle in other areas. While I don't say this from experience, it seems clear that administration and management of either of the free databases is far easier than Oracle. They also scale in a way that Oracle does not -- they scale down.
The OSS databases aren't going to beat Oracle at the reliability, enterprise, etc. They may be able to be good enough, which is what's important. They already have beaten Oracle at maintainability.
The same is true of Linux as well. Linux isn't more reliable or more powerful than the proprietary Unices. But it's far more manageable, easier to administer, and it scales down better than those other operating systems. Those other systems are dying... sure, they'll be around for a long time, but they've become a legacy. No doubt Oracle will be around for a very long time, but if PostgreSQL or MySQL can become good enough in Oracle's arena, then Oracle runs a serious risk of becoming a legacy database.
Pprior art for this should be even easier to find. NCSA httpd had server-side includes, which are primarily used for this purpose. But maybe they are specifically thinking of dynamic content, in which case specific CGI software, or software that created static HTML pages, would be the proper prior art.
So there, I can bitch about all ads, all the time if I want to! I can't do as much about the magazine ads, though...
Really, though, let's not pretend that ads in our real life aren't without their cost.
You are not allowed to use LGPL as a static library, without falling under the GPL. Some LGPLed software has specific exceptions to this (for instance, in places where it's expected to be used in an embedded environment). I believe, for instance, that Guile has this exception in its license that allows you to use it as a static library without falling under the GPL.
That's an easy situation. What about PHP? To make PHP install nicely, it has to know I'm using Apache, and how Apache is configured, and it has to make Apache know that .php means PHP should handle the file. On Debian it does this, all on its own. To do this without a package manager (and good policies), the OS either has to have strong policy (which means you can't just drag the application anywhere and use it), or it can have a registry, which is just a way of hiding the crud, and we all know how bad an idea that is. Or you're stuck. If PHP can't detect where Apache lives, you're stuck, your installation process will suck.
On a system with rich functionality, extensible software, and widely shared libraries and modules, drag-and-drop installation is a step backward. I have 1500 packages installed on my computer, that's only possible because of a great package management system, with developers backing it up.
Sure, Debian packages could stand to be better annotated. The average user doesn't care about all the libraries that are installed, and a good installer would hide these packages (for installation, selection, and upgrading). I assume Lindows does something like this with its service. The essential system underneith Debian is far more advanced than anything Windows or Mac has.
But after using it, I think it's mostly time wasted. The system's filesystem doesn't need to look pretty -- rather, the user shouldn't see it at all. When you go up directories as far as you can, you should not reach some overriding root directory in which everything is stored.
There's two very different things that a filesystem is used for -- user files, and application/system files. There's no reason for these to use the same metaphors or to be accessible through the same mechanisms.
Application files are usually useless out of context. They belong to the application, are updated at its discretion, and all operations on them are done by the application (perhaps at the user's request). We keep them in a hierarchical filesystem, keyed by strings separated with slashes, because that works well enough and it's quite fast (compared to an RDBMS, which would be silly, since its features would be wasted). "Applications" should be read to include the essential system software as well as end-user applications.
User files are something else entirely. They have nothing to do with application files, except that they are traditionally implemented the same way. User files aren't updated or controlled by the system, they aren't sorted by the system, they don't belong to the system or to an application (though they are manipulated through applications). They warrant the overhead of arbitrary metadata, versioning, and regular backups. RDBMS concepts could be usefully applied to these files.
Object oriented concept should also be applied, as opposed to the everything-is-a-stream-of-bytes metaphor that the traditional filesystem uses. This provides a basis upon which more complex data can be presented through the filesystem metaphor -- turning an mbox file into a folder, presenting pieces of hardware as files. There's no reason to bother actually presenting these as traditional files (like /hardware), because they are only usable with intermediary programs. Let those programs present these as files in this higher-level user filesystem.
The conclusion I gain from this is that this work doesn't need to be done on the filesystem level. Let package management systems deal with that -- they do so effectively. Instead the applications that a user interacts with need to present this totally new filesystem, which incidentally stores its information on the old filesystem (though every user file may actually be implemented as a directory, for instance). Gnome VFS is along these lines, and I believe KDE offers a similar abstraction, though neither is as ambitious as what I described... but maybe they'll continue to move in that direction. Nautilus had some of these ambitions as well, though I always felt it fell far short, even as it was bogged down by the features it did have.
Anyway, the New OS should happen in the desktop application arena. Traditional Unix (like Linux as implemented by Debian or Redhat) works well and is a good foundation. People just need the ambition to build new metaphors ontop of that foundation, not try to port the foundation to the new metaphor itself.
When Lindows started being bundled with cheap computers I was optimistic that this might happen -- but the reviews seem to agree that the hardware is underpowered and not very good quality.
Do you have any plans on offering higher-end computers (particularly higher quality), branded specifically for Lindows? Plans for branded peripherals that work seemlessly with that hardware? Computers that are aesthetically pleasing?