Which is, of course, the source of the problem. People running Linux boxes and using telnet are really not a lot different than people running IIS Index Server and not applying service packs. They're people content with the default setup, with a configuration that lets you do what you want, and they don't worry about anything past that.
If Linux/BSD/*nix becomes more mainstream, this won't change. They'll install Mandrake, or Red Hat, or Slackware, or Debian, or whatever, and as soon as it works, leave it alone. Telnet is weak and bind has a hole, but you can get SSH and a patch. NT has security holes; MS issues patches, hotfixes, and service packs. If you WANT to keep your system secure on either platform, you can. The problem isn't the inherent security of the platform, it's the willingness of people to deal with it.
From that perspective, NT even be could be considered better, because it makes it easier for lazy people to keep updated. Every once in a while MS releases a service pack or combined patch that fixes a whole bunch of security holes and bugs all at once. It's more likely that a lazy admin grabs the new ultrapatch once in a while, than it is that he checks to see if he's running bind x.xx or another of the many, many components that might have holes in them.
Any time a product or technology is used (or appears to be used) primarily for illict, illegal, or harmful purposes, it is the court's -responsibility- to look for legal uses of the product. This applies to anything. Guns, lockpicks, pepper spray, and peer-to-peer sharing. This doesn't imply that all technology is evil until proven good, but if something is obviously used illegally and never used for the Forces Of Good, then there's no reason to allow to to exist and plenty of reasons to ban it.
I don't support the banning of peer-to-peer. Not because I believe the ban will be unsuccessful. Bans are almost never 100% successful; that's not their point. Their point is to state that you can't ____, and give the justice system grounds to prosecute you if you do do it. I don't support the ban because I put P2P in the same category as baseball bats - easy to use for illegal purposes, but just as many legal uses.
But I object to Timothy's statement that "It's a sad state of things when you've got to prove that something is good". That's incorrect. It's a sad state when the courts DON'T try to prove something is good. They already know there are illegal uses. If the courts said "P2P is used for some illegal purposes, so ban it", they'd be dropping their responsibility. By saying "Are there legal uses, and are they persuasive enough that P2P should be legal?" they acknowledge that they've only seen the illegal uses and that that's probably a one-sided story.
Now, does anyone know if there's an immediate reason the EFF wants this list, or is it for future use? Either way I support it, but I'd like to know if the courts or legal teams have requested this information.
Then I'll watch what happens down south and ponder the ramifications this might have up here in Canada, where the entire topic has hardly come up.
You should be able to compress any individual piece of data that you choose. Of course the decompression script that you write will not necessarily work for all such data.
True enough...if you don't consider the decompressor part of the output. Similar to what Patrick did, I could check the first byte of the file, strip it, and write a "decompressor" that sticks the byte back. However, if you include the decompressor, then the decompressor is part of the output, and the "decompression algorithm" is actually "run this program through an x86 (or whatever)". Like it says (all over the place, here), you can't do it generically. That's not to say it can't be done for a given file...maybe I can find an LCG pseudo-random number generator that creates the first 50k of the file.
The technique which was used in this case would not work for instane if there was a massive under representation of 5's in the data set.
Of course, in that case, you could use the lack of fives to your advantage. Maybe a Huffman encoding?
Oh, don't get me wrong; I don't think communism will work. You've hit the point perfectly. Even without your comments on what you believe to be "right" or "wrong" (which a communist supporter would never accept), you're right in saying that one or more people that don't accept communism can and will screw the system up. I think Mark said that everyone would realize that communism benefitted everyone; I think that people would realize that if everyone BUT them was communist, they came out on top.
My point was (and is) that a lot of thinking which is common around Slashdotters and free-software advocates has a lot of fundamental ideas similar to communism. Consider these statements (from the FSF home page):
...you may find yourself using a proprietary program. If your friend asks to make a copy, it would be wrong to refuse.
The system of copyright gives software programs "owners"...(this is then construed as a Bad Thing)
...I am working to build a system...based on voluntary cooperation, and decentralization.
The Communist Manifesto mocks the modern concept of "Hard-won, self-acquired, self-earned property" as having no value, and Stallman similarly derides the statement "I put my sweat, my heart, my soul into this program. It comes from me, it's mine!" The FSF talks about "a system where people are free to decide their own actions", while The Marxist Society talks about "working to build a system where people are free to decide their own actions"...identical wording! Marx talks endlessly about the revolution of the social classes, and RMS discusses a revolution of today's software market.
This isn't saying that I support communism, or that I disdain the FSF. I think that, while communism may have some laudable goals, it will never work, and attempts to make it work do far more harm than good. I think that the FSF is an excellent organization, and I fully support it, but I find some of their views extremist. But that's not my point.
I remain convinced that if it were not for the massive American social stigma against communism, left especially from the Cold War, it would be an extremely prevalent viewpoint around Slashdot. It meshes perfectly with the views about patents and intellectual property that are so commonly expressed.
That old saying about "If you download MP3's, you are downloading communisim." is completely backwards. "If you don't download MP3's, you are promoting communisim." is more like it!
Actually, the saying is more correct than incorrect. Communism is not totalitarianism, and does not suggest an eternal ruling class. Rather, it is the idea that all people in a nation are equal and have equal rights to everything, regardless of who created it or where it is. "Ownership of property" is a meaningless concept in communism. All people will work equally and do what they're best at, to promote the well-being of everyone else.
Of course, the so-called "communist countries" don't actually implement communism; they're totalitarian. It seems that it's far too tempting to:
a) slack off, not work, and just ride on everyone else, and
b) take advantage of the government position and increase its influence, rather than wait for it to naturally dissolve, as Marx believed.
Now, I don't promote communism as a national policy, and I don't support what are popularly called "communist countries". But I think that if it weren't for the massive failure of places like Russia, a *lot* of Slashdotters/GPLers would be promoting a communist viewpoint. It's got a lot in common.
I think it's more likely that it would bring Windows people to Linux than Linux people to Windows. I've spoken to several people who are of the opinion that "Linux might be more stable, but it's hard to use and none of the programs I want run on it." If Office ran on Linux, I'd use a lot less of Windows. Microsoft knows this.
I will believe this when (if?) I see it. It wouldn't surprise me (much) if they did this after a corporate breakup, since the applications group wouldn't worry nearly as much about supporting the OS group. IE has already been ported to some Unix variants....But these rumours have been around many, many times in the past.
Just because I don't believe it's true, that doesn't mean I'm opposed to it. There are several posts up already of the "Oh no, it's the end of the world!" flavour. The existance of an application does not degrade an operating system, and may entice more people to put more development into Linux. Most people that decry this as the worst thing that can possibly happen say they are more concerned with Linux being a open-source, non-commercial, community platform than a successful operating system. So why is it that when Loki ports the latest and greatest game, they are hailed as everyone's heroes, but when MS ports a very good office suite, it's a disaster?
I love Linux as it is - with a free and open kernel, a command-line interface, choices between KDE, Gnome, or none...and I think that its fundamentals won't change if certain programs are released. By -law-, the kernel, XFree86, and bash will remain open source. If one group were to become "evil" in some fashion, another group would grab the code and split the project.
I don't think Office is coming anytime soon. But if it is, I applaud it. It is a -good- suite. Bloated in places, but very, very good. If the bloat were reduced, chances are a feature I wanted would be removed. If not, another one would be.
(Note that the exact mechanics of this vary from language to language, but the idea is the same)
Say you have an object in memory, 100 bytes big. Your program uses a reference to find the object in memory and do useful things to it, like change it. Now, if you copy it by reference, you make a new reference to the same object -- You have one object and two references. However, if you copy it by value, you make a new copy of the object, and a new reference pointing to the new object.
If you copied by reference, then the two references affect each other; make a change using reference A, and the change is reflected in reference B. It uses less memory (since there's only one copy of the data) and is a faster operation (there's less data to duplicate). However, garbage collection is difficult, because the object can only be safely deleted once ALL references are gone.
During a copy by value, the data is replicated, so making changes with reference A doesn't affect reference B. You need more memory (200 bytes -- there's now two copies of the original object) and copying the data takes time. However, garbage collection is very simple, because there's no possibility of two references to one object. Once you remove the reference, the object can be deleted.
This type of discussion is appropriate to a language like Java, where there are no pointers, only references to objects. In C or C++, you have variables (like "int num") and pointers ("int *numptr"). Pointers act as references and are used to copy objects by reference. It's a bit more complex, since pointers are often used to generate references to variables, so the same object is sometimes copied by value and sometimes by reference....
Hope this helps...I think I made sense:-) And everyone feel free to jump in if I messed anything up -- I despise Java from the pit of my being, so I don't know it that well. C is more my thing.
I agree in the fullest. I've seen a lot of GUIs, and a lot of them make life easier for simple tasks, but none of them let me do a "cp *.c *.h local; diff server local > patch;". This is an operation I perform all the time on a piece of software I work on (Maybe I have the syntax wrong, I put it in a shell script ). Some may say, "Yes, but [X] utility does that for you!" but I'd reply, "In Linux, I don't need [X] utility installed - it's part of the Gnu binutils and part of the shell". I run in console mode, not X...I like the speed and saved memory. I don't have a swapfile. If I'm in XWindows, I generally need one.
So the question seems to be...do we want a majority-user OS, or a programmer's/network-admin's OS? Or both? It seems obvious to me that since it's programmers and admins who use Linux now, it'll stay targetted at them.
Right now, Linux systems are highly customizable. I set up a Linux system on a 486/66 with 8MB of RAM on a 50MB hard drive as a MP3 player with no monitor or mouse and a minimal boot/shutdown time. It can still compile, it still runs bash, but there's a LOT missing. You can't do this in Windows. The tradeoff is that in Linux, you aren't guaranteed to get the same functionality from one setup to another. As a result, you see software advertising "Requires: RedHat 6.1". Perhaps we need a standardized "Home Linux Specification"? I still want the option to make a 50MB Slackware install, but a "Standard" installation option would be great.
I've talked to fellow Linux users about some glaring user-non-friendlynesses of Linux, and why Windows is better for many people - only to have them reply, "Yeah, but Linux is BETTER. If they just learned the OS and dumped Windows, everyone would be HAPPIER." Some people don't realize that people don't -want- to learn a tricky OS, even if it is faster, smaller, and stabler. They want to turn on their computer, surf the web, and write a letter to Aunt Nancy.
I consider myself a fairly experienced Linux user - I've been using it for a couple years now, and I started from a Slackware installation that's now quite customized. I'm running a local caching DNS server, and my network runs off a cable modem plugged into a Linux IP-Masquerading box. But it wasn't easy. Slackware's netconfig doesn't seem to support two network cards, so I had to set the second one up manually. It only took me half an hour, but it required a login as root to muck around in the config files. My system didn't let non-root users mount and unmount the CD-ROM and floppy, so I had to edit fstab. This kind of operation is trivial to me now, but in the beginning it took me an hour to find out that fstab existed, let alone what it did. To a neophyte computer user, it's essentially impossible.
Some of the best areas I see Linux improving in are: - Package management. RPMs are nice. Slackware's TGZs are nice. We need something standardized, however, and something that everyone uses. RPMs seem to be approaching that goal. This brings up the topic of: - Software installation. If it's not a precompiled binary, you need to compile it yourself. A "source package" would be nice - something like an RPM where it configures, compiles, and installs programs for you, and provides uninstall and "make clean" mechanisms. One of the best things this could do is resolve compile errors caused by libraries. Bringing me to: - Library management. My lib directory is apporaching the disaster of the DLLs in Windows\System. I've got libraries installed there without knowing what programs need them. There's symlinks galore to resolve problems with programs looking for old versions of the same library. As a whole, I'm not certain what I can do to that mass without bringing down installed software. I still haven't managed to get the latest Netscape Communicator installed, so it's a good thing I normally use Lynx. - Directory structure./usr/local/bin may seem like a nice place to keep your stuff, but installing everything to the same place gets messy fast. When programs are just one file (like a lot of Linux stuff is, fortunately), it's great. But as software gets bigger, it requires data files and configuration. Something akin to the "c:\Program Files" directory might clear this up, especially if/usr/local/bin then becomes a repository for symlinks to the actual binaries. Maybe something could be done with the mass of.* config files in my home directory too - a ".config" directory, in which each app creates its own directory? - Configuration. There's a -lot- of config files out there (/etc/* !) that need to be modified to customize your system. I know some programs out there provide a nice front-end to these, and I'm considering writing one myself, but I don't know of any that do it all. This is probably related to the various distributions putting certain key files (rc.*) in different places or under different names.
Now I'm not talking about hardware recognition and driver support. I'm not talking about installation. I'm not even talking about software availability. These are known issues, and issues that a lot of progress has been made on. I'm talking about maintenance of an existing system to stop it from becoming the administrative nightmare that it might be (Or, for that matter, that Windows 9x is, with system\ and the Registry). Some of these problems may already be resolved, and maybe solutions are being worked on for the others. If so, I'd love to hear (madhobbit@geocities.com).
I use Linux daily and love it. I'm also an extreme Computer Person. Linux needs a lot of work before even an unexperienced-but-willing-to-learn person has a good shot at it, and more yet before it's -really- mainstream. It's doing great in the network market...now let's see it hit the homes!
Slashdot Mirror
Which is, of course, the source of the problem. People running Linux boxes and using telnet are really not a lot different than people running IIS Index Server and not applying service packs. They're people content with the default setup, with a configuration that lets you do what you want, and they don't worry about anything past that.
If Linux/BSD/*nix becomes more mainstream, this won't change. They'll install Mandrake, or Red Hat, or Slackware, or Debian, or whatever, and as soon as it works, leave it alone. Telnet is weak and bind has a hole, but you can get SSH and a patch. NT has security holes; MS issues patches, hotfixes, and service packs. If you WANT to keep your system secure on either platform, you can. The problem isn't the inherent security of the platform, it's the willingness of people to deal with it.
From that perspective, NT even be could be considered better, because it makes it easier for lazy people to keep updated. Every once in a while MS releases a service pack or combined patch that fixes a whole bunch of security holes and bugs all at once. It's more likely that a lazy admin grabs the new ultrapatch once in a while, than it is that he checks to see if he's running bind x.xx or another of the many, many components that might have holes in them.
The Mad Hobbit
Any time a product or technology is used (or appears to be used) primarily for illict, illegal, or harmful purposes, it is the court's -responsibility- to look for legal uses of the product. This applies to anything. Guns, lockpicks, pepper spray, and peer-to-peer sharing. This doesn't imply that all technology is evil until proven good, but if something is obviously used illegally and never used for the Forces Of Good, then there's no reason to allow to to exist and plenty of reasons to ban it.
I don't support the banning of peer-to-peer. Not because I believe the ban will be unsuccessful. Bans are almost never 100% successful; that's not their point. Their point is to state that you can't ____, and give the justice system grounds to prosecute you if you do do it. I don't support the ban because I put P2P in the same category as baseball bats - easy to use for illegal purposes, but just as many legal uses.
But I object to Timothy's statement that "It's a sad state of things when you've got to prove that something is good". That's incorrect. It's a sad state when the courts DON'T try to prove something is good. They already know there are illegal uses. If the courts said "P2P is used for some illegal purposes, so ban it", they'd be dropping their responsibility. By saying "Are there legal uses, and are they persuasive enough that P2P should be legal?" they acknowledge that they've only seen the illegal uses and that that's probably a one-sided story.
Now, does anyone know if there's an immediate reason the EFF wants this list, or is it for future use? Either way I support it, but I'd like to know if the courts or legal teams have requested this information.
Then I'll watch what happens down south and ponder the ramifications this might have up here in Canada, where the entire topic has hardly come up.
The Mad Hobbit
True enough...if you don't consider the decompressor part of the output. Similar to what Patrick did, I could check the first byte of the file, strip it, and write a "decompressor" that sticks the byte back. However, if you include the decompressor, then the decompressor is part of the output, and the "decompression algorithm" is actually "run this program through an x86 (or whatever)". Like it says (all over the place, here), you can't do it generically. That's not to say it can't be done for a given file...maybe I can find an LCG pseudo-random number generator that creates the first 50k of the file.
The technique which was used in this case would not work for instane if there was a massive under representation of 5's in the data set.
Of course, in that case, you could use the lack of fives to your advantage. Maybe a Huffman encoding?
Oh, don't get me wrong; I don't think communism will work. You've hit the point perfectly. Even without your comments on what you believe to be "right" or "wrong" (which a communist supporter would never accept), you're right in saying that one or more people that don't accept communism can and will screw the system up. I think Mark said that everyone would realize that communism benefitted everyone; I think that people would realize that if everyone BUT them was communist, they came out on top.
...you may find yourself using a proprietary program. If your friend asks to make a copy, it would be wrong to refuse.
...I am working to build a system...based on voluntary cooperation, and decentralization.
My point was (and is) that a lot of thinking which is common around Slashdotters and free-software advocates has a lot of fundamental ideas similar to communism. Consider these statements (from the FSF home page):
The system of copyright gives software programs "owners"...(this is then construed as a Bad Thing)
The Communist Manifesto mocks the modern concept of "Hard-won, self-acquired, self-earned property" as having no value, and Stallman similarly derides the statement "I put my sweat, my heart, my soul into this program. It comes from me, it's mine!" The FSF talks about "a system where people are free to decide their own actions", while The Marxist Society talks about "working to build a system where people are free to decide their own actions"...identical wording! Marx talks endlessly about the revolution of the social classes, and RMS discusses a revolution of today's software market.
This isn't saying that I support communism, or that I disdain the FSF. I think that, while communism may have some laudable goals, it will never work, and attempts to make it work do far more harm than good. I think that the FSF is an excellent organization, and I fully support it, but I find some of their views extremist. But that's not my point.
I remain convinced that if it were not for the massive American social stigma against communism, left especially from the Cold War, it would be an extremely prevalent viewpoint around Slashdot. It meshes perfectly with the views about patents and intellectual property that are so commonly expressed.
That old saying about "If you download MP3's, you are downloading communisim." is completely backwards. "If you don't download MP3's, you are promoting communisim." is more like it! Actually, the saying is more correct than incorrect. Communism is not totalitarianism, and does not suggest an eternal ruling class. Rather, it is the idea that all people in a nation are equal and have equal rights to everything, regardless of who created it or where it is. "Ownership of property" is a meaningless concept in communism. All people will work equally and do what they're best at, to promote the well-being of everyone else. Of course, the so-called "communist countries" don't actually implement communism; they're totalitarian. It seems that it's far too tempting to:
a) slack off, not work, and just ride on everyone else, and
b) take advantage of the government position and increase its influence, rather than wait for it to naturally dissolve, as Marx believed.
Now, I don't promote communism as a national policy, and I don't support what are popularly called "communist countries". But I think that if it weren't for the massive failure of places like Russia, a *lot* of Slashdotters/GPLers would be promoting a communist viewpoint. It's got a lot in common.
I think it's more likely that it would bring Windows people to Linux than Linux people to Windows. I've spoken to several people who are of the opinion that "Linux might be more stable, but it's hard to use and none of the programs I want run on it." If Office ran on Linux, I'd use a lot less of Windows. Microsoft knows this.
I will believe this when (if?) I see it. It wouldn't surprise me (much) if they did this after a corporate breakup, since the applications group wouldn't worry nearly as much about supporting the OS group. IE has already been ported to some Unix variants....But these rumours have been around many, many times in the past.
Just because I don't believe it's true, that doesn't mean I'm opposed to it. There are several posts up already of the "Oh no, it's the end of the world!" flavour. The existance of an application does not degrade an operating system, and may entice more people to put more development into Linux. Most people that decry this as the worst thing that can possibly happen say they are more concerned with Linux being a open-source, non-commercial, community platform than a successful operating system. So why is it that when Loki ports the latest and greatest game, they are hailed as everyone's heroes, but when MS ports a very good office suite, it's a disaster?
I love Linux as it is - with a free and open kernel, a command-line interface, choices between KDE, Gnome, or none...and I think that its fundamentals won't change if certain programs are released. By -law-, the kernel, XFree86, and bash will remain open source. If one group were to become "evil" in some fashion, another group would grab the code and split the project.
I don't think Office is coming anytime soon. But if it is, I applaud it. It is a -good- suite. Bloated in places, but very, very good. If the bloat were reduced, chances are a feature I wanted would be removed. If not, another one would be.
The times, they are a-changin'
-- Bob Dylan
It's pretty simple...
:-) And everyone feel free to jump in if I messed anything up -- I despise Java from the pit of my being, so I don't know it that well. C is more my thing.
(Note that the exact mechanics of this vary from language to language, but the idea is the same)
Say you have an object in memory, 100 bytes big. Your program uses a reference to find the object in memory and do useful things to it, like change it. Now, if you copy it by reference, you make a new reference to the same object -- You have one object and two references. However, if you copy it by value, you make a new copy of the object, and a new reference pointing to the new object.
If you copied by reference, then the two references affect each other; make a change using reference A, and the change is reflected in reference B. It uses less memory (since there's only one copy of the data) and is a faster operation (there's less data to duplicate). However, garbage collection is difficult, because the object can only be safely deleted once ALL references are gone.
During a copy by value, the data is replicated, so making changes with reference A doesn't affect reference B. You need more memory (200 bytes -- there's now two copies of the original object) and copying the data takes time. However, garbage collection is very simple, because there's no possibility of two references to one object. Once you remove the reference, the object can be deleted.
This type of discussion is appropriate to a language like Java, where there are no pointers, only references to objects. In C or C++, you have variables (like "int num") and pointers ("int *numptr"). Pointers act as references and are used to copy objects by reference. It's a bit more complex, since pointers are often used to generate references to variables, so the same object is sometimes copied by value and sometimes by reference....
Hope this helps...I think I made sense
I agree in the fullest. I've seen a lot of GUIs, and a lot of them make life easier for simple tasks, but none of them let me do a "cp *.c *.h local; diff server local > patch;". This is an operation I perform all the time on a piece of software I work on (Maybe I have the syntax wrong, I put it in a shell script ). Some may say, "Yes, but [X] utility does that for you!" but I'd reply, "In Linux, I don't need [X] utility installed - it's part of the Gnu binutils and part of the shell". I run in console mode, not X...I like the speed and saved memory. I don't have a swapfile. If I'm in XWindows, I generally need one.
So the question seems to be...do we want a majority-user OS, or a programmer's/network-admin's OS? Or both? It seems obvious to me that since it's programmers and admins who use Linux now, it'll stay targetted at them.
Right now, Linux systems are highly customizable. I set up a Linux system on a 486/66 with 8MB of RAM on a 50MB hard drive as a MP3 player with no monitor or mouse and a minimal boot/shutdown time. It can still compile, it still runs bash, but there's a LOT missing. You can't do this in Windows. The tradeoff is that in Linux, you aren't guaranteed to get the same functionality from one setup to another. As a result, you see software advertising "Requires: RedHat 6.1". Perhaps we need a standardized "Home Linux Specification"? I still want the option to make a 50MB Slackware install, but a "Standard" installation option would be great.
I've talked to fellow Linux users about some glaring user-non-friendlynesses of Linux, and why Windows is better for many people - only to have them reply, "Yeah, but Linux is BETTER. If they just learned the OS and dumped Windows, everyone would be HAPPIER." Some people don't realize that people don't -want- to learn a tricky OS, even if it is faster, smaller, and stabler. They want to turn on their computer, surf the web, and write a letter to Aunt Nancy.
/usr/local/bin may seem like a nice place to keep your stuff, but installing everything to the same place gets messy fast. When programs are just one file (like a lot of Linux stuff is, fortunately), it's great. But as software gets bigger, it requires data files and configuration. Something akin to the "c:\Program Files" directory might clear this up, especially if /usr/local/bin then becomes a repository for symlinks to the actual binaries. Maybe something could be done with the mass of .* config files in my home directory too - a ".config" directory, in which each app creates its own directory?
I consider myself a fairly experienced Linux user - I've been using it for a couple years now, and I started from a Slackware installation that's now quite customized. I'm running a local caching DNS server, and my network runs off a cable modem plugged into a Linux IP-Masquerading box. But it wasn't easy. Slackware's netconfig doesn't seem to support two network cards, so I had to set the second one up manually. It only took me half an hour, but it required a login as root to muck around in the config files. My system didn't let non-root users mount and unmount the CD-ROM and floppy, so I had to edit fstab. This kind of operation is trivial to me now, but in the beginning it took me an hour to find out that fstab existed, let alone what it did. To a neophyte computer user, it's essentially impossible.
Some of the best areas I see Linux improving in are:
- Package management. RPMs are nice. Slackware's TGZs are nice. We need something standardized, however, and something that everyone uses. RPMs seem to be approaching that goal. This brings up the topic of:
- Software installation. If it's not a precompiled binary, you need to compile it yourself. A "source package" would be nice - something like an RPM where it configures, compiles, and installs programs for you, and provides uninstall and "make clean" mechanisms. One of the best things this could do is resolve compile errors caused by libraries. Bringing me to:
- Library management. My lib directory is apporaching the disaster of the DLLs in Windows\System. I've got libraries installed there without knowing what programs need them. There's symlinks galore to resolve problems with programs looking for old versions of the same library. As a whole, I'm not certain what I can do to that mass without bringing down installed software. I still haven't managed to get the latest Netscape Communicator installed, so it's a good thing I normally use Lynx.
- Directory structure.
- Configuration. There's a -lot- of config files out there (/etc/* !) that need to be modified to customize your system. I know some programs out there provide a nice front-end to these, and I'm considering writing one myself, but I don't know of any that do it all. This is probably related to the various distributions putting certain key files (rc.*) in different places or under different names.
Now I'm not talking about hardware recognition and driver support. I'm not talking about installation. I'm not even talking about software availability. These are known issues, and issues that a lot of progress has been made on. I'm talking about maintenance of an existing system to stop it from becoming the administrative nightmare that it might be (Or, for that matter, that Windows 9x is, with system\ and the Registry). Some of these problems may already be resolved, and maybe solutions are being worked on for the others. If so, I'd love to hear (madhobbit@geocities.com).
I use Linux daily and love it. I'm also an extreme Computer Person. Linux needs a lot of work before even an unexperienced-but-willing-to-learn person has a good shot at it, and more yet before it's -really- mainstream. It's doing great in the network market...now let's see it hit the homes!