A demo code to show what happens when pipe buffer get filled.

pipetest.py

"""This code runs a subprocess that produces a large amount of data, while
the main process reads the data at a slower pace. What happens on the long run?
This program is a demo showing what happens when a pipe buffer gets filled.
"""

import io
import sys
import time
import textwrap
import tempfile
import subprocess


if __name__ == '__main__':
    # Program to run, will write data both on stdout and stderr
    sub_program = textwrap.dedent('''
    import sys
    import time
    i = 0
    while True:
        print("I am still writing number", i, file=sys.stderr)
        print("Data" , i, "is data", i, 'filler ' * 20)
        i += 1
        time.sleep(0.01)  # Do not flood too much
    ''')

    # Open a temp file with source code
    with tempfile.NamedTemporaryFile(mode='wt') as program_file:
        # Write the source to the file
        print(sub_program, file=program_file)
        # Ensure code is there
        program_file.flush()
        
        # Start a python interpreter to run that code
        proc = subprocess.Popen(['python3', '-u', program_file.name],
                                stdout=subprocess.PIPE,
                                stderr=sys.stderr)
        # Read one line every second
        for line in io.TextIOWrapper(proc.stdout, encoding="utf-8"):
            print(line.strip())
            time.sleep(1)