Why do people switch to Linux 13489

On Sunday 30 October 2005 23:17, T.G. Reaper stood up and spoke the following words to the mbuttes incomp.os.linux.advocacy...:

The chances to fragmentation are severely reduced by the ability in GNU-Linux to spread out the system's directory structure over multiple filesystem, many of which you can mount read-only.

Both the spreading and the read-only facet are important in the reduction of filesystem fragmentation, as you are physically separating the static data from the dynamic data.

But yes, the risk to filesystem fragmentation does exist, albeit far less than in Windows, where the operating system itself - and if not specified otherwise by the user - the application software are installed in the same filesystem, with many of the applications generating lots of temporary files which do not respect the systemwide %TEMP and %TMP variables.

Been there, seen that.

The problem is that you Windows advocates make all kinds of accusations towards GNU-Linux in respect to its complexity and user-unfriendliness.

Windows is far less stable in this respect and thus requires more maintenance, but now you are saying that the user should maintain his system and should be paying attention?

Where is the argument of Windows beingeasierthan GNU-Linux now?

The above two lines are the only lines that compare Windows with VMS in regards to security, but they are not saying anything about true multi-user functionality.

Additionally, the last line above this paragraph should be completed with the words "when installed on a filesystem that supports them".

No I'm not - see my comments above.

To support them does not mean to require them. GNU-Linux supports the use of webcams, but that doesn't mean that a webcam should be plugged in in order for the system to be able to work properly.

NT can work properly without being installed on NTFS. This means that the NTFS security system through ACL's is an additional layer that is not required, and that therefore processes can be executed without that their ownership and permissions are checked. How would you go about and checking those if they aren't present?

If the above possibility to execute processes in such manner exists, then the layer of security can be bypbutted by malicious code. I'm sure you see the logic in that?

If you give all users permission to do everything on the filesystem, then you would indeed allow them to erase randomly selected files. Still, the core system would check the permissions before executing the command, as this is a required step in the end of code on the that matter.

In addition, there are some security precautions that cannot be bypbutted without that the user effectively has the root pbuttword, as some processes check the UID of the user starting the process. They may then either run with limited privileges or refuse to run at all.

You Windows advocates are often telling us that we are mentioning flaws in older Windows versions, but you bring up the 2.4 kernel generation?

No, the table does *not* show that. What the table shows is that the NT-based versions of Windows are designed to *support* ACL's, not to *require* them as part of their basic functionality. Hence, the core can do very well without them.

I would agree on that. SELinuxmakes heavy use of ACL's - aside from other extra precautions in the kernel.

Hmm... I wouldn't know about that. SGI'sXFSfilesystem has supported ACL's from the beginning, to my knowledge, albeit that it may not have been in GNU-Linux -XFSis the native filesystem of SGI IRIX.

Read my comments above, please. VMS was a multi-user system, yes, but that doesn't make Windows one. BeOS also used elements from the UNIX design, but was not a UNIX and certainly not a multi-user system. It didn't even have any file protection at all.

-- With kind regards,

*Aragorn* (Registered GNU-Linux user #223157)