fix(pty): drain late terminal query reply on teardown

[sweb]

Linked Issue

Closes #1257

Summary

Quitting an interactive TUI (e.g. Claude Code) under nono run left stray characters such as 3;1R in the next shell prompt.

3;1R is the tail of a terminal Cursor Position Report (ESC [ 3 ; 1 R), the reply to a ESC[6n query the TUI emits during shutdown. The proxy forwards the query to the terminal, but the reply round-trips back after the child has exited, so nothing on the PTY consumes it. release_terminal_for_prompt restores cooked mode with tcsetattr(..., TCSANOW, ...), which does not flush the input queue, so the stray reply is handed to the shell.

A bare tcflush does not help: the reply has usually not arrived at that instant. The fix waits for it, then flushes.

Agent Disclosure (if applicable)

This change was prepared with the assistance of an AI coding agent (Claude Code), reviewed, altered and applied by the human contributor.

Test Plan

• make fmt-check
• make clippy
• make test-cli
• Manual: nono run -- claude, /q, confirmed no 3;1R in the shell prompt.

Checklist

• An issue exists and is linked above
• All commits are signed-off, using DCO
• All new code follows the project's coding standards (CLAUDE.md) and is covered by tests
• Public-facing changes are paired with documentation updates
• Release note has been added to CHANGELOG.md if needed

[github-actions]

PR Review Summary

Size

Metric	Value
Lines added	+175
Lines removed	-2
Total changed	177
Classification	Medium (50–300 lines)

Affected crates

• crates/nono-cli — CLI changes. Verify argument parsing, flag documentation, and UX behaviour across supported platforms.

Blast radius — Contained

This PR touches: source code

---

_{Updated automatically on each push to this PR.}

[gemini-code-assist]

Code Review

This pull request introduces a mechanism to discard late terminal query replies (such as cursor-position reports) during final teardown to prevent them from leaking into the shell prompt. This is achieved by waiting briefly via poll and flushing the input queue using tcflush. Feedback is provided regarding a potential race condition where tcflush might be called before the entire multi-byte escape sequence has fully arrived, suggesting a brief sleep and a check on pfd.revents before flushing.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

When poll returns > 0, it indicates that at least one byte of the terminal query reply (such as the initial ESC character) has arrived in the input queue. However, because the terminal query reply is a multi-byte sequence (e.g., ESC[3;1R), calling tcflush immediately might only discard the first few bytes if the remaining bytes are still in transit or have not yet been processed by the OS tty layer. This race condition can result in the tail of the sequence (like 3;1R) still being left in the input queue and handed to the shell.

To prevent this, we should:

1. Verify that POLLIN is actually set in pfd.revents to ensure we only proceed when there is readable data.
2. Introduce a very brief sleep (e.g., 10ms) after poll succeeds to allow the entire multi-byte sequence to be fully received before flushing.

    if unsafe { libc::poll(&mut pfd, 1, timeout_ms) } > 0 && (pfd.revents & libc::POLLIN) != 0 {
        // Sleep briefly to ensure the entire multi-byte escape sequence has arrived in the input queue.
        std::thread::sleep(Duration::from_millis(10));
        // SAFETY: fd is the live terminal fd for the duration of the call.
        let tty = unsafe { std::os::fd::BorrowedFd::borrow_raw(fd) };
        let _ = nix::sys::termios::tcflush(tty, nix::sys::termios::FlushArg::TCIFLUSH);
    }

[sweb]

addressed in 6bff299