set -e is kinda useless

wat.sh

#!/usr/bin/env sh
# run this shell script with one of these shells:
# zsh, bash, dash, or (busybox) ash.
set -e

# false # this exits the shell immediately.
# false || false # this exits as well.
false && true # wait, what?
! true # huh?

fun() {
    false # this *should* cause, the shell to exit, right...?
    [ "$1" = 0 ]
}

if fun 0; then echo wat 1; fi
if false; then echo nope; elif fun 0; then echo wat 2; fi

i=0
while fun $i; do i=1; done
until fun $i; do i=0; done

# ! fun 0 # this exits zsh, and only zsh.
# (! (! fun 0)) # this exits bash, and only bash.
fun 0 && echo wat 3 || echo nope
fun 1 && echo nope
fun 0 || echo nope

echo aborting...
fun 0 # let's get out of here.
echo at least that works.

(set -e;$([ ];:))
if [ $? = 0 ]; then
    echo "You are running bash or busybox ash!"
fi

parser exploitation with squelched error message:

#!/usr/bin/env sh
set -e # you can remove this
echo This text is seen in any shell.
! { return 0;} 2>&- # exit any shell that isn't bash
echo This text is only seen in bash.

some shells (dash, busybox ash, yash) will exit the script with a nonzero value (1) so watch out for that

if (f(){ ! return;};f); then
    echo "You are running bash or mksh or ksh or posh!"
fi

subshell optional but then you're clobbering f()