Differences between revisions 1 and 12 (spanning 11 versions)
Revision 1 as of 2007-05-02 23:11:07
Size: 3986
Editor: redondos
Comment:
Revision 12 as of 2009-08-18 12:23:29
Size: 4009
Editor: localhost
Comment: Another formatting fix (indent the more-splitting reference)
Deletions are marked like this. Additions are marked like this.
Line 1: Line 1:
[[Anchor(faq24)]]
== I set variables in a loop. Why do they suddenly disappear after the loop terminates? ==
<<Anchor(faq24)>>
== I set variables in a loop. Why do they suddenly disappear after the loop terminates? Or, why can't I pipe data to read? ==
Line 4: Line 4:
The following command always prints "total number of lines: 0", although the variable {{{linecnt}}} has a larger value in the {{{while}}} loop: === The problem ===

In most shells, each command of a pipeline is executed in a separate SubShell.
Line 7: Line 9:
    # Non-working example (except in ksh88/ksh93)
Line 8: Line 11:
    cat /etc/passwd | while read line     printf "%s\n" foo bar | while read -r line
Line 10: Line 13:
        linecnt=`expr $linecnt + 1`         linecnt=$((linecnt+1))
Line 12: Line 15:
    echo "total number of lines: $linecnt"     echo "total number of lines: $linecnt" # prints 0

    # the problem also occurs without a loop
    var=0
    echo 2 | read -r var
    echo $var # also prints 0
Line 15: Line 23:
The reason for this surprising behaviour is that a {{{while/for/until}}} loop runs in a subshell when its input or output is redirected from a pipeline. For the {{{while}}} loop above, a new subshell with its own copy of the variable {{{linecnt}}} is created (initial value, taken from the parent shell: "0"). This copy then is used for counting. When the {{{while}}} loop is finished, the subshell copy is discarded, and the original variable {{{linecnt}}} of the parent (whose value has not changed) is used in the {{{echo}}} command. The reason for this surprising behaviour is that a {{{while/for/until}}} loop runs in a SubShell when it's part of a pipeline. For the {{{while}}} loop above, a new subshell with its own copy of the variable {{{linecnt}}} is created (initial value, taken from the parent shell: "0"). This copy then is used for counting. When the {{{while}}} loop is finished, the subshell copy is discarded, and the original variable {{{linecnt}}} of the parent (whose value has not changed) is used in the {{{echo}}} command.
Line 17: Line 25:
It's hard to tell when shell would create a new process for a loop:
 * BourneShell creates it when the input or output is redirected, either by using a pipeline or by a redirection operator ('<', '>').
 * ["BASH"] creates a new process only if the loop is part of a pipeline
 * KornShell creates it only if the loop is part of a pipeline, but ''not'' if the loop is the last part of it.
Different shells behave differently when using redirection or pipes with a loop:
 * BourneShell creates a subshell when the input or output of anything (loops, case etc..) but a simple command is redirected, either by using a pipeline or by a redirection operator ('<', '>').
 * [[BASH]] creates a new process only if the loop is part of a pipeline
 * KornShell creates it only if the loop is part of a pipeline, but ''not'' if the loop is the last part of it.  (The example above actually ''works'' in ksh88 and ksh93!)
 * POSIX specifies the bash behaviour, but as an extension allows any or all of the parts of the pipeline to run without a subshell (thus permitting the KornShell behaviour, as well).
Line 22: Line 31:
To solve this, either use a method that works without a subshell (shown below), or make sure you do all processing inside that subshell (a bit of a kludge, but easier to work with): === Workarounds ===
Line 24: Line 33:
{{{
    linecnt=0
    cat /etc/passwd |
    (
        while read line ; do
                linecnt="$((linecnt+1))"
        done
        echo "total number of lines: $linecnt"
    )
}}}
Several possibilities to avoid the subshell exist:
Line 35: Line 35:
To avoid the subshell completely (not easily possible if the other part of the pipe is a command!), use redirection, which does not have this problem at least for ["BASH"] and KornShell (but still for BourneShell):  * If the input is a file, remove the '''useless use of cat''':
Line 37: Line 37:
{{{
    linecnt=0
    while read line ; do
        linecnt="$((linecnt+1))"
   done < /etc/passwd
 {{{
  # POSIX
  while read -r line; do linecnt=$(($linecnt+1)); done < file
  echo $linecnt
 }}}

 Unfortunately this doesn't work with a Bourne shell; see [[http://heirloom.sourceforge.net/sh/sh.1.html#20|sh(1) from the Heirloom Bourne Shell]] for a workaround.

 * '''Group the commands''' and do it all in the subshell:

 {{{
  # POSIX
  linecnt=0
  cat /etc/passwd |
  {
      while read -r line ; do
         linecnt=$((linecnt+1))
      done
      echo "total number of lines: $linecnt"
  }
 }}}

 * Use ProcessSubstitution (BASH only):

 {{{
    # Bash
    while read -r line; do
       linecnt=$((linecnt+1))
    done < <(grep PATH /etc/profile)
    echo "total number of lines: $linecnt"
 }}}

 See also [[BashFAQ/001|FAQ #1]]

 * Use a [[NamedPipes|named pipe]] (POSIX):
 {{{
   # POSIX
   mkfifo mypipe
   grep PATH /etc/profile > mypipe &
   while read -r line;do
       linecnt=$(($linecnt+1))
   done < mypipe
Line 43: Line 80:
}}}  }}}
Line 45: Line 82:
For ["BASH"], when the first part of the pipe is a command, you can use "process substitution". The command used here is a simple "echo -e $'a\nb\nc'" as a substitute for a command with a multiline output:  * Use a '''coprocess''' (ksh, even pdksh, oksh, mksh..):
 {{{
  # ksh
  grep PATH /etc/profile |&
  while read -r -p line; do
    linecnt=$((linecnt+1))
  done
  echo "total number of lines: $linecnt"
 }}}
Line 47: Line 92:
{{{
    while read LINE; do
        echo "-> $LINE"
    done < <(echo -e $'a\nb\nc')
}}}
 * Another useful trick (using Bash/ksh93 syntax) is breaking a variable into words using {{{read}}}:
Line 53: Line 94:
A portable and common work-around is to redirect the input of the {{{read}}} command using {{{exec}}}:  {{{
  # Bash
  echo "$foo" | read -r a b c # this doesn't work
  read -r a b c <<< "$foo" # but this does
 }}}
Line 55: Line 100:
{{{
    linecnt=0
    exec < /etc/passwd # redirect standard input from the file /etc/passwd
    while read line # "read" gets its input from the file /etc/passwd
    do
        linecnt=`expr $linecnt + 1`
    done
    echo "total number of lines: $linecnt"
}}}
 Again, the pipeline causes the {{{read}}} command in the first example to run in a subshell, so its effect is never witnessed in the parent process. The second example does not create any subshells, so it works as we expect. The {{{<<<}}} operator is specific to bash (2.05b and later), and the input which follows it is usually called a "here string".
Line 65: Line 102:
This works as expected, and prints a line count for the file /etc/passwd. But the input is redirected from that file permanently. What if we need to read the original standard input sometime later again? In that case we have to save a copy of the original standard input file descriptor, which we later can restore:  For more examples of how to break input into words, see [[BashFAQ/001|FAQ #1]].
Line 67: Line 104:
{{{
    exec 3<&0 # save original standard input file descriptor "0" as FD "3"
    exec 0</etc/passwd # redirect standard input from the file /etc/passwd

    linecnt=0
    while read line # "read" gets its input from the file /etc/passwd
    do
        linecnt=`expr $linecnt + 1`
    done

    exec 0<&3 # restore saved standard input (fd 0) from file descriptor "3"
    exec 3<&- # close the no longer needed file descriptor "3"

    echo "total number of lines: $linecnt"
}}}

Subsequent {{{exec}}} commands can be combined into one line, which is interpreted left-to-right:

{{{
    exec 3<&0
    exec 0</etc/passwd
    _...read redirected standard input..._
    exec 0<&3
    exec 3<&-
}}}

is equivalent to

{{{
    exec 3<&0 0</etc/passwd
    _...read redirected standard input..._
    exec 0<&3 3<&-
}}}
 * With a POSIX shell you can use a here document instead:
 {{{
  # POSIX
  read -r a b c << EOF
  $foo
  EOF
 }}}

I set variables in a loop. Why do they suddenly disappear after the loop terminates? Or, why can't I pipe data to read?

The problem

In most shells, each command of a pipeline is executed in a separate SubShell.

    # Non-working example (except in ksh88/ksh93)
    linecnt=0
    printf "%s\n" foo bar  | while read -r line
    do
        linecnt=$((linecnt+1))
    done
    echo "total number of lines: $linecnt" # prints 0

    # the problem also occurs without a loop
    var=0
    echo 2 | read -r var
    echo $var # also prints 0

The reason for this surprising behaviour is that a while/for/until loop runs in a SubShell when it's part of a pipeline. For the while loop above, a new subshell with its own copy of the variable linecnt is created (initial value, taken from the parent shell: "0"). This copy then is used for counting. When the while loop is finished, the subshell copy is discarded, and the original variable linecnt of the parent (whose value has not changed) is used in the echo command.

Different shells behave differently when using redirection or pipes with a loop:

  • BourneShell creates a subshell when the input or output of anything (loops, case etc..) but a simple command is redirected, either by using a pipeline or by a redirection operator ('<', '>').

  • BASH creates a new process only if the loop is part of a pipeline

  • KornShell creates it only if the loop is part of a pipeline, but not if the loop is the last part of it. (The example above actually works in ksh88 and ksh93!)

  • POSIX specifies the bash behaviour, but as an extension allows any or all of the parts of the pipeline to run without a subshell (thus permitting the KornShell behaviour, as well).

Workarounds

Several possibilities to avoid the subshell exist:

  • If the input is a file, remove the useless use of cat:

      # POSIX
      while read -r line; do linecnt=$(($linecnt+1)); done < file
      echo $linecnt

    Unfortunately this doesn't work with a Bourne shell; see sh(1) from the Heirloom Bourne Shell for a workaround.

  • Group the commands and do it all in the subshell:

      # POSIX
      linecnt=0
      cat /etc/passwd |
      {
          while read -r line ; do
             linecnt=$((linecnt+1))
          done
          echo "total number of lines: $linecnt"
      }
  • Use ProcessSubstitution (BASH only):

        # Bash
        while read -r line; do
           linecnt=$((linecnt+1))
        done < <(grep PATH /etc/profile)
        echo "total number of lines: $linecnt"

    See also FAQ #1

  • Use a named pipe (POSIX):

       # POSIX
       mkfifo mypipe
       grep PATH /etc/profile > mypipe &
       while read -r line;do
           linecnt=$(($linecnt+1))
       done < mypipe
       echo "total number of lines: $linecnt"
  • Use a coprocess (ksh, even pdksh, oksh, mksh..):

      # ksh
      grep PATH /etc/profile |&
      while read -r -p line; do
        linecnt=$((linecnt+1))
      done
      echo "total number of lines: $linecnt"
  • Another useful trick (using Bash/ksh93 syntax) is breaking a variable into words using read:

      # Bash
      echo "$foo" | read -r a b c      # this doesn't work
      read -r a b c <<< "$foo"         # but this does

    Again, the pipeline causes the read command in the first example to run in a subshell, so its effect is never witnessed in the parent process. The second example does not create any subshells, so it works as we expect. The <<< operator is specific to bash (2.05b and later), and the input which follows it is usually called a "here string".

    For more examples of how to break input into words, see FAQ #1.

  • With a POSIX shell you can use a here document instead:
      # POSIX
      read -r a b c << EOF
      $foo
      EOF

BashFAQ/024 (last edited 2023-12-12 13:15:33 by 195)