path: root/chapter_05.xml

<?xml version="1.0"?>
<!DOCTYPE chapter PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
  "/usr/share/xml/docbook/xml-dtd-4.5/docbookx.dtd">

<chapter>
<title>The Bourne Again Shell</title>

<section>
<title>What Is A Shell?</title>

<para>
Yeah, what exactly is a shell?  Well, a shell is basically a
command-line user environment.  In essence, it is an application that
runs when the user logs in and allows him to run additional
applications.  In some ways it is very similar to a graphical user
interface, in that it provides a framework for executing commands and
launching programs.  There are many shells included with a full install
of Slackware, but in this book we're only going to discuss
<application>bash</application>(1), the Bourne Again Shell. Advanced
users might want to consider using the powerful
<application>zsh</application>(1), and users familiar with older UNIX
systems might appreciate <application>ksh</application>. The truly
masochistic might choose the <application>csh</application>, but new
users should stick to <application>bash</application>.
</para>

</section>

<section>
<title>Environment Variables</title>

<para>
All shells make certain tasks easier for the user by keeping track of
things in environment variables.  An environment variable is simply a
shorter name for some bit of information that the user wishes to store
and make use of later.  For example, the environment variable PS1 tells
<application>bash</application> how to format its prompt.  Other
variables may tell applications how to run.  For example, the LESSOPEN
variable tells <application>less</application> to run that handy
<filename>lesspipe.sh</filename> preprocessor we talked about, and
LS_OPTIONS tuns on color for <application>ls</application>.
</para>

<para>
Setting your own envirtonment variables is easy.
<application>bash</application> includes two built-in functions for
handling this: <application>set</application> and
<application>export</application>.  Additionally, an environment
variable can be removed by using <application>unset</application>.
(Don't panic if you accidently unset an environment variable and don't
know what it would do.  You can reset all the default variables by
logging out of your terminal and logging back in.)  You can reference a
variable by placing a dollar sign ($) in front of it.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>set FOO=bar</userinput>
<prompt>darkstar:~$ </prompt><userinput>echo $FOO</userinput>
bar
</screen>

<para>
The primary difference between <application>set</application> and
<application>export</application> is that
<application>export</application> will (naturally) export the variable
to any sub-shells.  (A sub-shell is simply another shell running inside
a parent shell.)  You can easily see this behavior when working with
the PS1 variable that controls the <application>bash</application>
prompt.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>set PS1='FOO '</userinput>
<prompt>darkstar:~$ </prompt><userinput>export PS1='FOO '</userinput>
<prompt>FOO </prompt>
</screen>

<para>
There are many important environment variables that
<application>bash</application> and other shells use, but one of the
most important ones you will run across is PATH.  PATH is simply a list
of directories to search through for applications.  For example,
<application>top</application>(1) is located at
<application>/usr/bin/top</application>.  You could run it simply by
specifying the complete path to it, but if
<filename>/usr/bin</filename> is in your PATH variable,
<application>bash</application> will check there if you don't specify a
complete path one your own.  You will most likely first notice this
when you attempt to run a program that is not in your PATH as a normal
user, for instance, <application>ifconfig</application>(8).
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>ifconfig</userinput>
bash: ifconfig: command not found
<prompt>darkstar:~$ </prompt><userinput>echo $PATH</userinput>
/usr/local/bin:/usr/bin:/bin:/usr/X11R6/bin:/usr/games:/opt/www/htdig/bin:.
</screen>

<para>
Above, you see a typical PATH for a mortal user.  You can change it on
your own the same as any other environment variable.  If you login as
root however, you'll see that root has a different PATH.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>su -</userinput>
Password: 
<prompt>darkstar:~# </prompt><userinput>echo $PATH</userinput>
/usr/local/sbin:/usr/sbin:/sbin:/usr/local/bin:/usr/bin:/bin:/usr/X11R6/bin:/usr/games:/opt/www/htdig/bin
</screen>

</section>

<section>
<title>Wildcards</title>

<para>
Wildcards are special characters that tell the shell to match certain
criteria.  If you have experience with DOS, you'll recognize * as a
wildcard that matches anything.  <application>bash</application> makes
use of this wildcard and several others to enable you to easily define
exactly what you want to do.
</para>

<para>
This first and most common of these is, of course, *.  The asterisk
matches any character or combination of characters, including none.
Thus <userinput>b*</userinput> would match any files named b, ba, bab,
babc, bcdb, and so forth.  Slightly less common is the ?.  This
wildcard matches one instance of any character, so
<userinput>b?</userinput> would match ba and bb, but not b or bab.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>touch b ba bab</userinput>
<prompt>darkstar:~$ </prompt><userinput>ls *</userinput>
b ba bab
<prompt>darkstar:~$ </prompt><userinput>ls b?</userinput>
ba
</screen>

<para>
No, the fun doesn't stop there!  In addition to these two we also have
the bracket pair "[ ]" which allows us to fine tune exactly what we
want to match.  Whenever <application>bash</application> see the
bracket pair, it substitues the contents of the bracket.  Any
combination of letters or numbers may be specified in the bracket as
long as they are comma seperacted.  Additionally, ranges of numbers and
letters may be specified as well.  This is probably best shown by
example.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>ls a[1-4,9]</userinput>
a1 a2 a3 a4 a9
</screen>

<para>
Since Linux is case-sensitive, capital and lower-case letters are
treated differently.  All capital letters come before all lower-case
letters in "alphabetical" order, so when using ranges of capital and
lower-case letters, make sure to get them right.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>ls 1[W-b]</userinput>
1W 1X 1Y 1Z 1a 1b
<prompt>darkstar:~$ </prompt><userinput>ls 1[w-B]</userinput>
/bin/ls: cannot access 1[b-W]: No such file or directory
</screen>

<para>
In the second example, 1[b-W] isn't a valid range, so the shell treats
it as a filename, and since that file doesn't exist,
<application>ls</application> tells you so.
</para>

</section>

<section>
<title>Tab Completion</title>

<para>
Still think there's entirely too much work involved with using
wildcards?  You're right.  There's an even easier way when you're
dealing with long filenames: tab completion.  Tab completion enables
you to type just enough of the filename to uniquely identify it, then
by hitting the TAB key, <application>bash</application> will fill in
the rest for you.  Even if you haven't typed in enough text to uniquely
identify a filename, the shell will fill in as much as it can for you.
Hitting TAB a second time will make it display a list of all possible
matches for you.
</para>

</section>

<section>
<title>Input and Output Redirection</title>

<para>
One of the defining features of Linux and other UNIX-like operating
systems is the number of small, relatively simple applications and the
ability to stack them together to create complex systems.  This is
achieved by redirecting the output of one program to another, or by
drawing input from a file or second program.
</para>

<para>
To get started, we're going to show you how to redirect the output of a
program to a file.  This is easily done with the '>' character.  When
<application>bash</application> sees the '>' character, it redirects
all of the standard output (also known as stdout) to whatever file name
follows.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>echo foo</userinput>
foo
<prompt>darkstar:~$ </prompt><userinput>echo foo > /tmp/bar</userinput>
<prompt>darkstar:~$ </prompt><userinput>cat /tmp/bar</userinput>
foo
</screen>

<para>
In this example, we show you what <application>echo</application> would
do if its stdout was not redirected to a file, then we re-direct it to
the <filename>/tmp/bar</filename> file.  If <filename>/tmp/bar</filename>
does not exist, it is created and the output from
<application>echo</application> is placed within it.  If
<filename>/tmp/bar</filename> did exist, then its contents are
over-written.  This might not be the best idea if you want to keep
those contents in place.  Thankfully, <application>bash</application>
supports '>>' which will append the output to the file.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>echo foo</userinput>
foo
<prompt>darkstar:~$ </prompt><userinput>echo foo > /tmp/bar</userinput>
<prompt>darkstar:~$ </prompt><userinput>cat /tmp/bar</userinput>
foo
<prompt>darkstar:~$ </prompt><userinput>echo foo2 >> /tmp/bar</userinput>
<prompt>darkstar:~$ </prompt><userinput>cat /tmp/bar</userinput>
foo
foo2
</screen>

<para>
You can also re-direct the standard error (or stderr) to a file.  This
is slightly different in that you must use '2>' instead of just '>'.
(Since <application>bash</application> can re-direct input, stdout, and
stderr, each must be uniquely identifiable.  0 is input, 1 is stdout,
and 2 is stderr.  Unless one of these is specified,
<application>bash</application> will make its best guess as to what you
actually meant, and assumed anytime you use '>' you only want to
redirect stdout.  1> would have worked just as well.)
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>rm bar</userinput>
rm: cannot remove `bar': No such file or directory
<prompt>darkstar:~$ </prompt><userinput>rm bar 2> /tmp/foo</userinput>
<prompt>darkstar:~$ </prompt><userinput>cat /tmp/foo</userinput>
rm: cannot remove `bar': No such file or directory
</screen>

<para>
You may also redirect the standard input (known as stdin) with the
'&lt;'
character, though it's not used very often.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>fromdos &lt; dosfile </userinput>
</screen>

<para>
Finally, you can actually redirect the output of one program as input
to another.  This is perhaps the most useful feature of
<application>bash</application> and other shells, and is accomplished
using the '|' character.  (This character is referred to as 'pipe'.
If you here some one talk of piping one program to another, this is
exactly what they mean.)
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>ps auxw | grep getty</userinput>
root      2632  0.0  0.0   1656   532 tty2     Ss+  Feb21   0:00 /sbin/agetty 38400 tty2 linux
root      3199  0.0  0.0   1656   528 tty3     Ss+  Feb15   0:00 /sbin/agetty 38400 tty3 linux
root      3200  0.0  0.0   1656   532 tty4     Ss+  Feb15   0:00 /sbin/agetty 38400 tty4 linux
root      3201  0.0  0.0   1656   532 tty5     Ss+  Feb15   0:00 /sbin/agetty 38400 tty5 linux
root      3202  0.0  0.0   1660   536 tty6     Ss+  Feb15   0:00 /sbin/agetty 38400 tty6 linux
</screen>

</section>

<section>
<title>Task Management</title>

<para>
<application>bash</application> has yet another cool feature to offer,
the ability to suspend and resume tasks.  This allows you to
temporarily halt a running process, perform some other task, then
resume it or optionally make it run in the background. Upon pressing
<keycap>CTRL-Z</keycap>, <application>bash</application> will suspend
the running process and return you to a prompt. You can return to that
process later.  Additionally, you can suspend multiple processes in
this way indefinitely.  The <application>jobs</application> built-in
command will display a list of suspended tasks.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>jobs</userinput>
[1]-  Stopped                 vi TODO
[2]+  Stopped                 vi chapter_05.xml
</screen>

<para>
In order to return to a suspended task, run the
<application>fg</application> built-in to bring the the most recently
suspended task back into the fore ground. If you have mutiple suspended 
tasks, you can specify a number as well to bring one of them to the
foreground.
</para>

<screen><prompt>darkstar:~$ </prompt><userinput>fg # "vi TODO"</userinput>
<prompt>darkstar:~$ </prompt><userinput>fg 1 # "vi chapter_05.xml"</userinput>
</screen>

<para>
You can also background a task with (surprize)
<application>bg</application>. This will allow the process to continue
running without maintaining control of your shell. You can bring it
back to the fore ground with <application>fg</application> in the same
way as suspended tasks.
</para>

</section>

<section>
<title>Terminals</title>

<para>
Slackware Linux and other UNIX-like operating systems allow users to
interact with them in many ways, but the most common, and arguably the
most useful, is the terminal. In the old days, terminals were keyboards
and monitors (sometimes even mice) wired into a mainframe or server via
serial connections. Today however, most terminals are virtual; that is,
they exist only in software.  Virtual terminals allow users to connect
to the computer without requiring expensive and often incompatabile
hardware. Rather, a user needs only to run the software and they are
presented with a (usually) highly customizable virtual terminal.
</para>

<para>
The most common virtual terminals (in that every Slackware Linux machine
is going to have at least one) are the gettys.
<application>agetty</application>(8) runs six instances by default on
Slackware, and allows local users (those who can physically sit down in
front of the computer and type at the keyboard) to login and run
applications. Each of these gettys is available on different tty
devices that are accessible seperately by pressing the
<keycap>ALT</keycap> key and one of the function keys from
<keycap>F1</keycap> through <keycap>F6</keycap>. Using these gettys
allows you to login multiple times, perhaps as different users, and run
applications in those users' shells silmutaneously. This is most
commonly done with servers which do not have
<application>X</application> installed, but can be done on any machine.
</para>

<para>
On desktops, laptops, and other workstations where the user prefers a
graphical interface provided by <application>X</application>, most
terminals are graphical.  Slackware includes many different graphical
terminals, but the most commonly used are KDE's
<application>konsole</application> and XFCE's
<application>Terminal</application>(1) as well as the old standby,
xterm(1). If you are using a graphical interface, check your tool bars
or menus. Each desktop environment or window manager has a virtual
terminal (often called a terminal emulater), and they are all labelled
differently. Typically though, you will find them under a "System"
sub-menu in desktop environments. Executing any of these will give you
a graphical terminal and automatically run your default shell.
</para>

</section>

<section>
<title>Customization</title>

<para>
By now you should be pretty familiar with
<application>bash</application> and you may have even noticed some odd
behavior. For example, when you login at the console, you're presented
with a prompt that looks a bit like this.
</para>

<screen><prompt>alan@darkstar:~$ </prompt></screen>

<para>
However, sometimes you'll see a much less helpful prompt like this one.
</para>

<screen><prompt>bash-3.1$ </prompt></screen>

<para>
The cause here is a special environment variable that controls the
<application>bash</application> prompt. Some shells are considered
"login" shells and others are "interactive" shells, and both types read
different configuration files when started. Login shells read
<filename>/etc/profile</filename> and
<filename>~/.bash_profile</filename> when executed. Interactive shells
read <filename>~/.bashrc</filename> instead. This has some advantages
for power users, but is a common annoyance for many new users who want
the same environment anytime they execute
<application>bash</application> and don't care about the difference
between login and interactive shells. If this applies to you, simply
edit your own ~/.bashrc file and include the following lines.
(For more information on
the different configuration files used, read the INVOCATION section of
the <application>bash</application> man page.)
</para>

<screen>
# ~/.bashrc
. /etc/profile
. ~/.bash_profile
</screen>

<para>
When using the above, all your login and interactive shells will have
the same environment settings and behave identically. Now, anytime we
wish to customize a shell setting, we only have to edit
<filename>~/.bash_profile</filename> for user-specific changes and
<filename>/etc/profile</filename> for global settings. Let's start by
configuring the prompt.
</para>

<para>
<application>bash</application> prompts come in all shapes, colors, and
sizes, and every user has their own preferances. Personally, I prefer
short and simple prompts that take up a minimum of space, but I've seen
and used mutli-line prompts many times. One personal friend of mine
even included ASCII-art in his bash prompt. To change your prompt you
need only to change your PS1 variable. By default, Slackware attempts
to configure your PS1 variable thusly:
</para>

<screen><prompt>darkstar:~$ </prompt>echo $PS1
\u@\h:\w\$ </screen>

<para>
Yes, this tiny piece of funny-looking figures controls your
<application>bash</application> prompt. Basicaly, every character in
the PS1 variable is included in the prompt, unless it is a escaped by a
<keycap>\</keycap>, which tells <application>bash</application> to
interpret it. There are many different escape sequences and we can't
discuss them all, but I'll explain these.  The first "\u" translates to
the username of the current user.  "\h" is the hostname of the machine
the terminal is attached to. "\w" is the current working directory, and
"\$" displays either a <keycap>#</keycap> or a <keycap>$</keycap> sign,
depending on whether or not the current user is root.  A complete
listing of all prompt escape sequences is listed in the
<application>bash</application> man page under the PROMPTING section.
</para>

<para>
Since we've gone through all this trouble to discuss the default
prompt, I thought I'd take some time to show you a couple example
prompts and the PS1 variable values needed to use them.
</para>

<screen><prompt>Wed Jan 14 12:08 AM
alan@raven:~$ </prompt>echo $PS1
\d \@\n\u@\h:\w$ 
<prompt>HOST: raven - JOBS: 0 - TTY: 3
alan@~/Desktop/sb_3.0:$ </prompt>echo $PS1
HOST: \H - JOBS: \j - TTY: \l\n\u@\w:\$
</screen>

<para>
For even more information on configuring your bash prompt, including
information on setting up colored prompts, refer to
<filename>/usr/doc/Linux-HOWTOs/Bash-Prompt-HOWTO</filename>. After
reading that for a short while, you'll get an idea of just how powerful
your <application>bash</application> prompts can be. I once even had a
prompt that gave me up to date weather information such as temperature
and barometric pressure!
</para>

</section>

</chapter>