Let's see.  We can discuss the nature of Unix filesystems.  This is sort of a
condensed discussion, feel free to ask questions.

I assume you are familiar with the directory tree structure of filesystems
that both Unix and DOS/Windows share.  A directory has subdirectories, and
subdirectories can have other subdirectories, and any of those directories
can have files.

In DOS and Windows, every disk has its own root directory, and each disk is
identified with a drive letter, C:, A:, etc.  Unix is not like this.  On any
Unix system, there is only one root directory.  Other disks are added into
the filesystem as subdirectories.  The root directory of a disk is "mounted"
as a subdirectory of another disk.

You can use the "df" command (display filesystems) to see how several disks
(or disk partitions) are connected into the overall filesystem.

	fury.coreth.com:/home/mpearce 829> df
	Filesystem  1K-blocks     Used    Avail Capacity  Mounted on
	/dev/ad0s1a    128990    34494    84178    29%    /
	/dev/ad0s1g   4562302  2431421  1765897    58%    /home
	/dev/ad0s1e   2064302  1226074   673084    65%    /usr
	/dev/ad0s1f    516062     8158   466620     2%    /var
	procfs              4        4        0   100%    /proc
	/dev/acd0c     650312   650312        0   100%    /cdrom

This command shows several columns.  The first column is the "device" node
associated with the disk or disk partition that has been mounted.  The last
column is its "mount point", the directory it has been connected at.

The root filesystem is on /dev/ad0s1a, which the system knows is the first
partition on the first slice of the first IDE hard disk drive.  If I change
to the root directory ("cd /"), I will be in a directory located on that
partition.  I have a CD-ROM mounted in the /cdrom directory.  If I change
to the /cdrom directory, I will be looking at files in the root directory
of the CD-ROM.

Unlike DOS and Windows, a filesystem must be mounted before it can be used,
and dismounted before it can be removed.  Part of the shutdown procedure is
to dismount all the filesystems.  Filesystems are mounted with the "mount"
command, and dismounted with the "umount" (not "unmount") command.  The
long way around mounting a CD-ROM, for example is:

	mount -t cd9660 /dev/acd0c /cdrom

This mounts the filesystem located in the CD-ROM drive (/dev/acd0c) into the
/cdrom directory.  The "-t cd9660" tells the mount command what type of
filesystem it is (ISO 9660 is the normal format for CD-ROM's).

Strictly speaking, a directory is a list of names and inode numbers.  The
inode number refers to a table on the disk containing information about a
file, including the type of file, the file's date, permissions, location on
the disk, etc.

In Unix, it is possible to have more than one name reference the same inode
number (and therefore the same file).  It can be two different names in the
same directory, or names in two different directories.  This is called a
"link" (that term predates HTML or the web).

There are something like seven types of files (let me see if I can remember
them).  There are regular files.  There are directories (a directory is
really a file containing names and inode numbers).  There are device nodes
(which can be character devices or block devices), FIFOs or named pipes, 
there are Unix domain sockets, and there are symbolic links.

In Unix, most devices have "filenames", and this filename is the device node.
Take the CD-ROM drive for an example.  There is a device driver in the system
for IDE (ATAPI) CD-ROM drives.  This driver knows how to control multiple
CD-ROM drives.  If a program wants to "open" or use the device driver to
control the hardware device, it needs to know its name.  A device node is
basically a cross reference between a name that can be used by a program, and
the device driver in the kernel.

Device nodes are created with the "mknod" command.  You need to supply the
type of device (block or character), the major and minor device numbers,
and the name you want to create.  Most device nodes are created for you when
the operating system is installed.  Knowing the necessary information to
create a device node that isn't there tends to be a challenging task.

A FIFO and a named pipe are the same thing.  A Unix domain socket is very
similar to a named pipe, except that it uses the same API as the TCP/IP
network socket library.  These are all used for inter-process communication.
Data sent into a pipe by one program or process can be read from the pipe
by another program or process.

A symbolic link is much like the link described above, but it is implemented
differently.  It is a small file referencing the real filename of some file.
It can be used in circumstances that normal ("hard") links cannot be used.
For instance, a hard link cannot be used with directory names.  A hard link
also cannot be used to create names for a file on two different filesystems.
A symbolic link is slower and uses more resources, but it is usually more
flexible.  Links (both hard and symbolic) are created with the "ln" command.
Symbolic links are often refered to as "symlinks".

All files (directories, device nodes, etc) have permissions, including an
owner and a group.  Every file is owned by some person.  It is also owned
by some group.  There are file permission bits that indicate what the owner,
the group, and everyone else can do to the file.  This information can be
seen with the "ls -l" command:

	fury.coreth.com:/home/mpearce 836> ls -l 1874.jpg
	-rw-r--r--  1 mpearce  staff  22031 Aug  5 12:07 1874.jpg

This file is owned by mpearce and the group staff.  The first column shows
permission bits.  There are ten positions.  The first position shows the file
type: '-' for a regular file, 'd' for a directory, etc.  The next three
positions show the read, write, and execute permissions for the owner of the
file.  An 'r' indicates that the owner can read the file.  A 'w' means that
the owner can write to (modify) the file.  An 'x' means that the owner can
execute the file as if it were a program.

After the three owner permission bits are three bits for the group.  They
have a similar meaning, and apply to any user in the staff group.  The last
three positions are bits indicating the permissions given to every elese on
the system.  If a user is not mpearce and not a member of the staff group,
then it is these last three bits that dictate what that user can do the file.

In the file above, anyone can read the file (it is often said to be "world
readable"), but only mpearce can modify the file.

File permissions are set with the "chmod" command.  Also, the owner or group
can be set with the "chown" and "chgrp" commands respectively.  Obviously,
only the owner of the file (and superuser) can set the permissions of the
file.

You will often see Unix people refer to file permissions in octal, a base
eight numbering system with digits 0 through 7.  I'm not really going to go
into that right now.  Some people seem to think it is easier to use the
octal numbers rather than the symbolic representation of the file
permissions.  I'm not sure why this is, unless they like to memorize a set of
common file permissions.  In the example above, -rw-r--r-- is the octal
number 644.  A lot of this is discussed more on the chmod manpage.

That ought to be enough to chew on.  Please let me know which parts don't
make sense.  Your reading assignment:
	man df
	man mount
	man umount
	man mknod
	man ln
	man ls
	man chmod
	man chown
	man chgrp