Shell Programming and Scripting

This reads single keystrokes and produces an output:

Code:

#! /bin/bash

while : ; do
    read -s -n 1 >/dev/null 2>&1
    echo ${REPLY}
done | awk '{print}'

This second one don't. Even though these examples make no sense; the real code is more complicated. Who knows what the problem is here? Something with STDIN and STDOUT?

Code:

#! /bin/bash

while : ; do
    read -s -n 1 >/dev/null 2>&1
    echo ${REPLY}
done | awk '{print}' | while read x ; do echo ${x} ; done

This is because your input never gets terminated. Meanwhile awk is buffering the output. Watch what happens when you replace the loop by:

Code:

while read -s -n 1 >/dev/null 2>&1; do
  case $REPLY in 
     "z") break
  esac
  echo ${REPLY}
done | awk '{print}' | while read x ; do echo ${x} ; done

Now type input characters and then enter a z . You'll notice that the letters get printed once you enter the z, which will terminate the loop en thus the output and the buffers get flushed.

Note that you first example worked, perhaps because you were using a specific awk (gawk?) and it seems to buffer its output only when it is writing to a file or pipe. When I tried it with mawk, you first example did not work because mawk was buffering there as well.

Buffer control is not standard in awk.
When using gawk, this should produce interactive output:

Code:

gawk '{print;fflush("")}'

When using mawk:

Code:

mawk -W interactive '{print}'

yap - this works. Thank you very much!