• expr is useful for pattern matching and for counting number of characters in string. For example:

      expr $string : 'xxx\(.*\)'  ## picks out stuff matched within ()
      expr $string : '.*'         ## length of string
  

  It can also be used for arithmetic

      VAR=`expr $VAR + 1`   # increments VAR
  

  although ksh's $(( )) is more convenient:

      VAR=$(( VAR + 1 ))
  

• To get rid of blanks padding a string use: STRING=$( echo $STRING )
• Instead of using LOGNAME to get a user name try following:

      USER=$( id | sed -e 's/.*(//' -e 's/).*//' )
      ID=$( id | sed -e 's/uid=//' -e 's/(.*//' )
  

• To handle terminals, use tput. For example, to output a word in bold:

      BOLD=`tput smso`
      BOLDOFF=`tput rmso`
      echo "This is ${BOLD}portable${BOLDOFF}."
  

• While it is not possible for a child to change a parent's environment, it is possible to make changes if both processes cooperate. For example, if the child is an awk, the child can generate suitable variable assignments and the parent can evaluate these:

       eval "`awk -f program`"  ## where awk program produces output like...
                                ##  VAR1=something
                                ##  VAR2=something_else
  

• Note that in variable assignments such as ${PAGER:=more} the colon : signifies that the empty string is to be treated the same as an unset string (in this case the default more is assigned in either case).

• Scripts should start #!/usr/bin/*sh not #!/bin/*sh
• It not usually a good idea to give full path names for commands in a shell, except perhaps for ls and rm which users may have altered in undesirable ways. If you rely on the PATH variable instead the script is more likely to be portable.
• Useful filters:
  xargs - for breaking long argument lists.
  cut - useful in some situations instead of sed.

Last update: 2000 December 10