gistfile1.txt

$ cat test.sh
#!/usr/bin/env bash

set -e

foo () {
    echo $(echo "subshell foo"; exit 11)
}

bar () {
    result=$(echo "subshell bar"; exit 22)
    echo $result
}

indirection () {
    var=$(bar)
    echo "indirect", $var
}

foo
indirection
bar
$ ./test.sh
subshell foo
indirect, subshell bar
$ echo $?
22
$

Today I Learned that set -e doesn't apply to a to a subshell unless you assign the result to a variable...

http://unix.stackexchange.com/questions/23026/how-can-i-get-bash-to-exit-on-backtick-failure-in-a-similar-way-to-pipefail

The issue seems to be with command substitutions specifically, and variable assignments having odd semantics around them.

 -e  Exit immediately if a command exits with a non-zero status.

$(...) is a command substitution, not a command. The actual command in foo is echo $(...), which exist with status 0. bar's command is a variable set without a command name:

If there is a command name left after expansion, execution proceeds as described below. Otherwise, the command exits. If one of the expansions contained a command substitution, the exit status of the command is the exit status of the last command substitution performed. If there were no command substitutions, the command exits with a status of zero.

It also looks like -e doesn't correctly propagate in bash, but does in dash. dash otherwise has the same semantics regarding the substitution.

$ bash
bash$ (set -e; echo $(echo foo; false; echo bar)))
foo bar
bash$ exit
$ sh
dash$ (set -e; echo $(echo foo; false; echo bar)))
foo

Well, at least this explains a lot of the similar issues I had in the past.

Also looks like github's having issues updating comments, the one with the link didn't show for me until I added my (now rather redundant) comment...