jdx/mise · error

cannot check staged changes in

Error message

cannot check staged changes in {}: {}

What it means

staged_paths runs `git diff --name-only --cached --` (with GIT_OPTIONAL_LOCKS=0) inside each relevant root to enumerate staged changes. If that git invocation exits non-zero, mise bails with this error including the root path and git's trimmed stderr. It surfaces underlying git failures (corrupt repo, permission problems, dubious ownership) that would otherwise be hidden.

Solutions

  1. Read the embedded git stderr in the message and fix that specific git error.
  2. For 'dubious ownership' errors, run `git config --global --add safe.directory <root>` for the reported path.
  3. Repair or remove the broken .git directory in the affected root, then retry the sync.
  4. Check permissions on the directory and ensure the user running mise can execute git there.

Example fix

// before
$ mise dot sync
error: cannot check staged changes in ~/.config: fatal: detected dubious ownership in repository at '/home/user/.config'
// after
$ git config --global --add safe.directory /home/user/.config
$ mise dot sync
Defensive patterns

Strategy: try-catch

Validate before calling

# pre-flight the exact git call mise makes
git -C "$ROOT" diff --name-only --cached -- >/dev/null || echo "git fails in $ROOT: $?"

Try / catch

match run_sync() {
  Err(e) if e.starts_with("cannot check staged changes in") => {
    // parse embedded git stderr, fix repo (safe.directory, permissions), retry
  }
  other => other,
}

Prevention

When it happens

Trigger: During `mise dot sync` pull, the per-root `git diff --cached` subprocess returns a non-zero status; the git stderr text is embedded in the message. Common causes: the directory is a broken/partial git repo, `safe.directory` dubious-ownership refusal, or filesystem permission errors running git.

Common situations: A dotfiles directory that was copied from another user/machine triggering git's 'dubious ownership' check; a corrupted .git directory in the work root; read-only mounts that prevent git from running.

Understand the failure class

Background: "git command failed": what it means when a tool shells out to git and git exits non-zero — this error's family across 21 libraries.

Related errors

AI-assisted analysis of jdx/mise@533346cc37 (2026-09-17). Data as JSON: /api/errors/9b2cb46d243e0d3d. Report an issue: GitHub.

Appendix: source

Thrown at src/system/history/sync/apply.rs:1025

            .arg("-C")
            .arg(&root)
            .args([
                "-c",
                "core.fsmonitor=false",
                "diff",
                "--cached",
                "--name-only",
                "-z",
                "--no-renames",
                "--no-ext-diff",
                "--no-textconv",
                "--",
            ])
            .env("GIT_OPTIONAL_LOCKS", "0")
            .stdin(std::process::Stdio::null())
            .output()?;
        if !output.status.success() {
            bail!(
                "cannot check staged changes in {}: {}",
                display_path(&root),
                String::from_utf8_lossy(&output.stderr).trim()
            );
        }
        for name in output
            .stdout
            .split(|byte| *byte == 0)
            .filter(|name| !name.is_empty())
        {
            #[cfg(unix)]
            let relative = {
                use std::os::unix::ffi::OsStrExt;
                PathBuf::from(std::ffi::OsStr::from_bytes(name))
            };
            #[cfg(not(unix))]
            let relative = PathBuf::from(std::str::from_utf8(name)?);
            staged.insert(normalize_target(&root.join(relative)));

View on GitHub (pinned to 533346cc37)