Confused about determining if a shell script is running interactively

Question

I only want to determine from my POSIX shell script, if it is running interactively, but for some reason, the following function:

running_interactively()
{
    printf '%s' ${-} | grep -F i > /dev/null 2>&1
}

returns false even if I run the script in terminal. Am I doing it wrong, or is the definition of interactive script somehow different from my plain idea of running the script by a user in terminal?

---

Snippet of the code:

#!/bin/sh

set -u

running_interactively()
{
#   echo $- returns only u
    printf '%s' ${-} | grep i > /dev/null 2>&1
}

print_error_and_exit()
{
    # redirect all output from this function to standard error stream
    if running_interactively
    then
        exec >&2
    else
        echo wrong again, smart ass
    fi
...
}

print_error_and_exit someArgs

Answer 1

A shell script is, unless it's sourced by an interactive shell, very seldom run in an interactive shell environment. This means that $- would not include an i.

What you could check is to see whether standard input is connected to a terminal or not. This is done using the -t test with an argument of 0 (the file descriptor of the standard input stream):

running_interactively () { [ -t 0 ]; }

This assumes that by "is running interactively" you mean "able to read input directly from a terminal".

An additional test on file descriptor 2 (standard error) would also be possible as a test of being able to do full interaction with the user in a script. User interaction mainly happens on standard input (user input) and standard error (prompts, diagnostic messages, etc.):

running_interactively () { [ -t 0 ] && [ -t 2 ]; }

However, testing on file descriptor 1 (standard output) would fail if the output of the script was redirected or piped.