What do the bash-builtins 'set' and 'export' do?

Question

I am at a bit of a loss as to the purpose of set and export in Bash (and I guess probably other shells too).

I would think it is for setting environment variables, but that can be done just with VARIABLE=VALUE, right?

Also typing set and export on their own show different values.

So what is their purpose?

Answer 1

export marks the given variable as exported to children of the current process, by default they are not exported. For example:

$ foo=bar
$ echo "$foo"
bar
$ bash -c 'echo "$foo"'

$ export foo
$ bash -c 'echo "$foo"'
bar

set, on the other hand, sets shell attributes and the positional parameters.

$ set foo=baz
$ echo "$1"
foo=baz

Note that baz is not assigned to foo, it simply becomes a literal positional parameter. There are many other things set can do (mostly shell options), see help set.

As for printing, export called with no arguments prints all of the variables in the shell's environment. set also prints variables that are not exported. It can also export some other objects (although you should note that this is not portable), see help export.

Answer 2

See help set or info bash set: set is used to set shell attributes and positional parameters.

Variables that are not exported are not inherited by child processes. export is used to mark a variable for export.

Answer 3

you can think of the 'export' command as turning the variable global...

for better understanding I would append to the previous answers the following example:

x=1
bash -c 'echo $x'
# nothing show

export x
bash -c 'echo $x'
# value of x shown

• for windows users the use of 'set' is confusing
  because the command is 'accepted' (gives no error)