feat(headless): persist evaluated commands to history database by eitsupi · Pull Request #134 · eitsupi/arf

eitsupi · 2026-03-29T07:29:52Z

Summary

Initialize SqliteBackedHistory in headless mode so that IPC evaluate and user_input commands are recorded in the same SQLite database used by the REPL
Generate session ID and pass history_session_id to the IPC server, matching REPL behavior
History saving is best-effort (failures logged, never block IPC responses); history.disabled config is respected
Add --history-dir and --no-history CLI flags to the headless subcommand, matching the REPL
Populate metadata fields (timestamp, hostname, cwd) on history items
IPC reply is sent before history persistence to avoid blocking clients on SQLite I/O
history_session_id is only advertised when the database was actually opened

Known limitations

start_timestamp is post-execution: The timestamp is set after evaluate_with_capture returns, not before. This matches REPL behavior (reedline sets the timestamp at save time), so the two modes are consistent.
hostname is fetched per-command: Could be cached since it doesn't change during a session. The syscall cost is negligible, so this is left as a future optimization.
HEADLESS_HISTORY is OnceLock: Cannot be re-initialized within the same process. Fine for production (one headless session per process), but limits in-process unit testing.

Test plan

cargo check / cargo clippy pass (including --tests)
All existing tests pass (unit + pty_ipc_tests + headless_tests)
New integration test: test_headless_history_persistence — verifies command_line, exit_status, hostname, cwd are saved correctly for evaluate (success/error) and user_input
New integration test: test_headless_no_history_flag — verifies --no-history prevents database creation
Shell completion snapshots updated

🤖 Generated with Claude Code

Initialize SqliteBackedHistory in headless mode so that IPC evaluate and user_input commands are recorded in the same SQLite database used by the REPL. The session ID and history_session_id are now generated and passed to the IPC server, matching the REPL behavior. History saving is best-effort: failures are logged but never block IPC response delivery. The history.disabled config setting is respected. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Copilot

Pull request overview

Enables command history persistence for headless mode so IPC-driven evaluate / user_input calls are recorded in the same SQLite history database used by the interactive REPL, and exposes a headless history_session_id via IPC session metadata for history isolation.

Changes:

Initialize a SqliteBackedHistory in headless mode and store it for IPC request handling.
Persist headless evaluate and user_input commands to the history DB (best-effort, log on failure).
Generate and pass history_session_id into IPC server startup in headless mode.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
crates/arf-console/src/main.rs	Adds headless history initialization and passes `history_session_id` to IPC server startup.
crates/arf-console/src/ipc/mod.rs	Adds a global headless history backend and saves IPC-evaluated commands into SQLite history.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

crates/arf-console/src/ipc/mod.rs

crates/arf-console/src/main.rs

crates/arf-console/src/ipc/mod.rs

HistoryItem::from_command_line() leaves all metadata fields as None. Set start_timestamp, hostname, and cwd so headless history entries match the REPL behavior and display correctly in the history browser. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

- Send IPC reply before saving to history so SQLite I/O does not delay the client response - Only advertise history_session_id when the SQLite backend was actually opened successfully, avoiding a misleading session JSON Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

into_string().ok() silently drops the hostname on systems with non-UTF-8 hostnames, while the REPL uses to_string_lossy() which always succeeds. Align the headless path to match. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Expose the same history configuration options available in the REPL (--history-dir, --no-history) for the headless subcommand. These override the config file settings, making it easy to control history from the command line and simplifying testing. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Copilot

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

crates/arf-console/src/ipc/mod.rs

crates/arf-console/src/main.rs

Verify that headless mode correctly saves evaluated commands to the SQLite history database with proper metadata (exit_status, hostname, cwd). Also verify that --no-history prevents database creation. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

…ry_dir Idiomatic Rust: &PathBuf is unnecessary double indirection since it derefs to &Path. Aligns with the other path parameters in the same function signature. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

The readiness probe only runs in polling mode (--quiet/--json/--log-file). This test uses stderr-based readiness, so no probe row is recorded. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Remove the claim that history_session_id is always None in headless mode, since headless now supports history persistence. The field is None only when history is disabled or the database failed to open. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Use Vec<_> with type inference instead of spelling out the full tuple type, which clippy flags as too complex. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Copilot

Pull request overview

Copilot reviewed 9 out of 9 changed files in this pull request and generated 5 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-29T08:04:42Z

crates/arf-console/src/ipc/mod.rs

+/// Once set, `headless_handle_request` will persist evaluated commands
+/// (both `evaluate` and `user_input`) to the SQLite history database.
+pub fn set_headless_history(history: reedline::SqliteBackedHistory) {
+    let _ = HEADLESS_HISTORY.set(Mutex::new(history));


set_headless_history ignores the result of OnceLock::set. If this function is accidentally called twice (e.g. in future refactors or tests that run headless initialization more than once), the second call will silently do nothing and keep using the old backend. Consider returning a Result or at least logging a warning when the history backend was already initialized.

Suggested change

let _ = HEADLESS_HISTORY.set(Mutex::new(history));

if HEADLESS_HISTORY

.set(Mutex::new(history))

.is_err()

{

log::warn!(

"Headless history backend already initialized; ignoring duplicate set_headless_history call"

);

}

Copilot · 2026-03-29T08:04:42Z

crates/arf-console/src/ipc/mod.rs

+            if !code.is_empty() {
+                let exit_status = if has_error { 1 } else { 0 };
+                save_to_headless_history(&code, Some(exit_status));


Headless history saving currently skips only strictly empty strings. In the REPL loop, whitespace-only input is ignored via let trimmed = line.trim(); if trimmed.is_empty() { continue; } (see crates/arf-console/src/repl/mod.rs:638-641). To keep behavior consistent and avoid saving blank/whitespace commands from IPC clients, use a trimmed emptiness check here as well.

Copilot · 2026-03-29T08:04:42Z

crates/arf-console/src/ipc/mod.rs

+            if !code.is_empty() {
+                save_to_headless_history(&code, Some(exit_status));
+            }


Same as the Evaluate path: this saves whitespace-only user_input to history because it checks !code.is_empty() rather than trimming. The REPL ignores whitespace-only input before evaluation/history (crates/arf-console/src/repl/mod.rs:638-641), so consider matching that behavior here too.

Copilot · 2026-03-29T08:04:43Z

crates/arf-console/tests/headless_tests.rs

+    // Small delay to let SQLite flush
+    std::thread::sleep(Duration::from_millis(200));
+


The test uses a fixed sleep(Duration::from_millis(200)) to wait for history persistence. This is timing-dependent and can be flaky on slow/loaded CI machines. Prefer polling for the DB file/expected rows with a bounded timeout (similar to other loops in this test file) so the test is deterministic.

Copilot · 2026-03-29T08:04:43Z

crates/arf-console/tests/headless_tests.rs

+///
+/// Verifies that:
+/// - IPC evaluate commands are saved with correct command_line and exit_status
+/// - Metadata fields (hostname, cwd, start_timestamp) are populated


This doc comment says the test verifies start_timestamp is populated, but the query/assertions only check command_line, exit_status, hostname, and cwd. Either include start_timestamp in the SELECT + assert it is non-null, or update the comment so it matches what the test actually validates.

Suggested change

/// - Metadata fields (hostname, cwd, start_timestamp) are populated

/// - Metadata fields (hostname, cwd) are populated

## Summary - Warn when `set_headless_history` is called more than once instead of silently ignoring the duplicate - Skip whitespace-only commands in headless history (Evaluate and UserInput paths) to match REPL behavior - Fix doc comment in test that incorrectly mentioned `start_timestamp` verification Addresses review feedback from Copilot on PR #134. Co-authored-by: Claude Opus 4.6 (1M context) <[email protected]>

eitsupi requested a review from Copilot March 29, 2026 07:30

Copilot started reviewing on behalf of eitsupi March 29, 2026 07:30 View session

Copilot AI reviewed Mar 29, 2026

View reviewed changes

crates/arf-console/src/ipc/mod.rs Show resolved Hide resolved

crates/arf-console/src/main.rs Outdated Show resolved Hide resolved

crates/arf-console/src/ipc/mod.rs Outdated Show resolved Hide resolved

eitsupi marked this pull request as draft March 29, 2026 07:35

eitsupi and others added 2 commits March 29, 2026 07:36

eitsupi requested a review from Copilot March 29, 2026 07:38

Copilot started reviewing on behalf of eitsupi March 29, 2026 07:38 View session

Copilot AI reviewed Mar 29, 2026

View reviewed changes

crates/arf-console/src/ipc/mod.rs Show resolved Hide resolved

crates/arf-console/src/main.rs Show resolved Hide resolved

eitsupi and others added 4 commits March 29, 2026 07:43

fix(test): remove incorrect comment about readiness probe in history

3aac86d

The readiness probe only runs in polling mode (--quiet/--json/--log-file). This test uses stderr-based readiness, so no probe row is recorded. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

eitsupi requested a review from Copilot March 29, 2026 07:48

fix(test): avoid clippy type_complexity warning in history test

573e6e9

Use Vec<_> with type inference instead of spelling out the full tuple type, which clippy flags as too complex. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

eitsupi marked this pull request as ready for review March 29, 2026 07:58

eitsupi merged commit 4e74999 into main Mar 29, 2026
10 checks passed

eitsupi deleted the feat/headless-history branch March 29, 2026 07:59

Copilot started reviewing on behalf of eitsupi March 29, 2026 08:00 View session

Copilot AI reviewed Mar 29, 2026

View reviewed changes

eitsupi mentioned this pull request Mar 29, 2026

fix(headless): address review feedback on history persistence #135

Merged

3 tasks

eitsupi mentioned this pull request Mar 29, 2026

chore: prepare release 0.2.6 #139

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(headless): persist evaluated commands to history database#134

feat(headless): persist evaluated commands to history database#134
eitsupi merged 10 commits intomainfrom
feat/headless-history

eitsupi commented Mar 29, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 29, 2026

Uh oh!

Copilot AI Mar 29, 2026

Uh oh!

Copilot AI Mar 29, 2026

Uh oh!

Copilot AI Mar 29, 2026

Uh oh!

Copilot AI Mar 29, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

-    let _ = HEADLESS_HISTORY.set(Mutex::new(history));
+    if HEADLESS_HISTORY
+        .set(Mutex::new(history))
+        .is_err()
+    {
+        log::warn!(
+            "Headless history backend already initialized; ignoring duplicate set_headless_history call"
+        );
+    }

		// Small delay to let SQLite flush
		std::thread::sleep(Duration::from_millis(200));

	/// - Metadata fields (hostname, cwd, start_timestamp) are populated
	/// - Metadata fields (hostname, cwd) are populated

Conversation

eitsupi commented Mar 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Known limitations

Test plan

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 29, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 29, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 29, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 29, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 29, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

eitsupi commented Mar 29, 2026 •

edited

Loading