Consider the following example:
file="test" echo $file
My idea is that file persists only during the command echo $file. Although $file does not persist after this command, echo prints nothing, as if $file were not set.
Is it possible to set one-command-lifelong variables in the shell?
edit: I don't want to resort to shell scripts or functions as suggested in all these related questions
The variable binding in your command line doesn't take place until the command line is executed. But the parameter substitution of $file takes place while the command is expanded, which must happen before it is executed.
The way around this is to put your command into a function:
fun()
{
echo "file = " $file
}
file=abc fun
var=value, POSIX leaves it unspecified whether or not the variable assignments persist after the completion of the function. In practice, it doesn't in ksh.
The command
file="test" echo $file
doesn't output test because the expansion occurs before running the command. Check this out:
$ set -x
$ foo='hello' echo $foo
+ foo=hello
+ echo
The same thing doesn't occur when you run a Unix shell script instead of running a shell command, e.g.,
foo='hello' /path/to/script.sh
In this case, the foo='hello' assignment is first resolved, then /path/to/script.sh is run.
Some previous and related posts proposed solutions for persistent variable assignment + CLI commands:
foo='hello' bash -c 'echo $foo': Using shell script is not desirable! Wordy approach to simple commands.(){ local var=value; echo "1: $var"; }; echo "2: $var": Using functions is not desirable! Wordy approach to simple commands.TEST=foo && echo $TEST: Simple but persistent.My way out is
(file="test"; echo $file)
By creating a subshell, you can create a shell variable, use it in the echo command, and then it won't persist outside this command.
foo is initially unset, after foo=hello bash -c 'echo $foo' it remains unset.
var=value command is exactly what you want: it sets the variable var to value only for the execution environment of command. Your solution here actually does make it persist outside, it's just that you are adding the extra step of running a subshell and the variable is set only in the subshell. So once the subshell exits, you no longer have the variable set, but that isn't what your question asked for.
file='test' eval 'echo $test' has the advantage that it does not create an extra shell.
(foo=xyz; echo "$foo"); echo "$foo" the modified value of foo effectively persists only for the duration of the first echo (including the expansions in that command), not the rest of the script (the second echo), so IMO, it looks to fill their desire exactly. Too bad something like foo=xyz { whatever... } doesn't work.
fileis set only forecho, but$fileis expanded beforeechoruns (it has to be) so$fileis empty when it’s expanded. Try the example given in the linked answer, using a shell script: you’ll see that the variable is indeed set for the command (in a non-persistent manner).file="test" echo $filedoes exactly what you want it to do. Can you edit and clarify the use case? Do you really want the variable to only be available in the command's environment, or do you want it to be available in the parent (which is whatecho "$var"requires) but to then be unset after the command finishes? If you only want it to exist in the command's environment, "for the lifetime of the command" thenvar=foo commandis the right way.