Curses programming: cursor handled incorrectly

[thautwarm]

Environment

Windows build number: Microsoft Windows [Version 10.0.19041.1]
Windows Terminal version : Version: 0.7.3451.0

CPython: 3.7.4
CPython package curses(Windows-curses)

Steps to reproduce

Open cmd or PowerShell inside Windows Terminal, with python run

def main():
    window = curses.initscr()
    curses.noecho()
    curses.nonl()
    curses.nocbreak()
    curses.raw()
    window.keypad(False)
    x = 0
    y = 0
    while True:
        window.refresh()
        c = window.getch()
        window.addstr(0, 0, str(c))
        if c == ord('q'):
            return
        elif c == ord('a'):
            x = 10
        elif c == ord('b'):
            y = 10
        else:
            x = 0
            y = 0
            pass

        window.move(y, x)
try:
    main()
finally:
    curses.endwin()

Expected behavior

Press a, the cursor move to (0, 10) immediately.

Actual behavior

After pressing a, the cursor doesn't move to (0, 10) immediately, i.e., the screen show doesn't change at all. Then after an another pressing, the cursor moves to (0, 10).

P.S:

• Expected behavior is according to the package curses bundled in Python of Linux distros.
• Things work perfectly when we use cmd or PowerShell without Windows Terminal.

zephyrproject-rtos/windows-curses#7

[zadjii-msft]

Yep that looks broken to me. Definitely something wrong in conpty, since this repros in inception (cmd.exe /c wsl.exe cmd.exe) as well. Thanks for the clear report with easy repro steps!

(future me: I put the file in %~%\dev\tmp)

[j4james]

I expect this is the same bug as #4107.

[DHowett-MSFT]

Nah, this one is separate from #4107. It repros in inception (#4102 (comment)), and the inbox conhost servicing interop hasn't been updated to the changeset that regressed 4107.

[ghost]

🎉This issue was addressed in #4372, which has now been successfully released as Windows Terminal Preview v0.9.433.0.:tada:

Handy links:

• Release Notes
• Store Download