I suppose you're correct about the terminating newline case, in that the newline is only added once. I'm not sure what you'd call it, but I'm talking about it

I suppose you're correct about the terminating newline case, in that the newline is only added once. I'm not sure what you'd call it, but I'm talking about it as if it were a kind of idempotent 'base case'. For example: If you do echo -en "" | ( (vim - -esbnN -c 'w!/dev/stderr|q!' >/dev/null) 2>&1) | xxd versus echo -en "" | ( (vim - -esbnN -c 'w!/dev/stderr|q!' >/dev/null) 2>&1) | ( (vim - -esbnN -c 'w!/dev/stderr|q!' >/dev/null) 2>&1) | xxd These will the same output, so it is idempotent. However, if you build upon that and do echo -en "" | ( (vim - -esbnN -c '%!xxd' -c 'w!/dev/stderr|q!' >/dev/null) 2>&1) | xxd versus echo -en "" | ( (vim - -esbnN -c '%!xxd' -c 'w!/dev/stderr|q!' >/dev/null) 2>&1) | ( (vim - -esbnN -c '%!xxd' -c 'w!/dev/stderr|q!' >/dev/null) 2>&1) | xxd The result is not the same, and therefore not idempotent if you plan to add extra -c commands.