Have a look at this little script:
#!/bin/bash
function do_something() {(
    set -e
    mkdir "/opt/some_folder"                                     # <== returns 1 -> abort?
    echo "mkdir returned $?"                                     # <== sets $0 to 0 again
    rsync $( readlink -f "${BASH_SOURCE[0]}" ) /opt/some_folder/ # <== returns 23 -> abort?
    echo "rsync returned $?"                                     # <== sets $0 to 0 again
)}
# here  every command inside `do_something` will be executed - regardless of errors
echo "run do_something in if-context.."
if ! do_something ; then
  echo "running do_something did not work"
fi
# here `do_something` aborts on first error
echo "run do_something standalone.."
do_something
echo $?
I was trying to do what was suggested here (don't miss the extra parentheses introducing a sub-shell) but I didn't execute the function (do_something in my case) separately but together with the if-expression.
Now when I run if ! do_something the set -e command seems to have no effect.
Can someone explain this to me?
 
     
     
    