In a shell script, by default, the next line is executed if an error occurs in the current line. Sometimes, we may want to stop running a script further after an error has been encountered. The -e option of set ensures to exit a script once any of the commands in a pipeline fails.

In the following shell script, do_not_exit_on_failure.sh doesn't use set with the option -e:

$ cat do_not_exit_on_failure.sh

#!/bin/bash
# Filename: do_not_exit_on_failure.sh
# Description: Resume script after an error

echo "Before error"
cd /root/       # Will give error
echo "After error"

After running this script, the output is as follows:

Before error
do_not_exit_on_failure.sh: line 6: cd: /root/: Permission denied
After error

We see that the command after the error gets executed as well. In order to stop the execution after an error is encountered, use set -e in the script. The following script demonstrates the same:

$ cat exit_on_failure.sh

#!/bin/bash
# Filename: exit_on_failure.sh
# Description: Exits script after an error

set -e
echo "Before error"
cd /root/       # Will give error
echo "After error"

The output after running the preceding script is as follows:

Before error
exit_on_failure.sh: line 7: cd: /root/: Permission denied

We can see that the script has been terminated after encountering an error at the line number 7.

Using set with the -P option doesn't resolve symbolic links. Following example demonstrate how we can enable or disable symbolic link resolution of /bin directory which is symbolic link of /usr/bin/ directory:

$ ls -l /bin
lrwxrwxrwx. 1 root root 7 Nov 18 18:03 /bin -> usr/bin
$ set –P    # -P enable symbolic link resolution
$ cd /bin
$ pwd
/usr/bin
$ set +P   # Disable symbolic link resolution
$ pwd
/bin

We can use the set command to see all local variables accessible for the current process. The local variables are not accessible in the subprocess.

We can create our own variable and set it locally as follows:

$ MYVAR="Linux Shell Scripting"
$ echo $MYVAR
 Linux Shell Scripting
$ set | grep MYVAR  # MYVAR local variable is created
MYVAR='Linux Shell Scripting'
$ bash    # Creating a new bash sub-process in current bash
$ set | grep MYVAR
$    # Blank because MYVAR is local variable

To make a variable accessible to its subprocesses as well, use the export command followed by the variable to be exported:

$ MYVARIABLE="Hello World"
$ export  MYVARIABLE
$ bash    # Creating a new bash sub-process under bash
$ echo $MYVARIABLE
Hello World

This will export the MYVARIABLE variable to any subprocess that ran from that process. To check whether MYVARIABLE has exported or not, run the following command:

$ export |grep MYVARIABLE
declare -x MYVARIABLE="Hello World"
$ export | grep MYVAR
$MYVAR variable is not present in sub-process but variable MYVARIABLE is present in sub-process.

To unset local or exported variables, use the unset command and it will reset the value of the variable to null:

$ unset MYVAR        # Unsets local variable MYVAR
$ unset  MYVARIABLE    # Unsets exported variable MYVARIABLE