Fix _logId() collision in SyncRepository with per-instance counter

[Copilot]

SyncRepository._logId() used only microsecondsSinceEpoch, allowing ID collisions when multiple logs are created within the same microsecond (e.g. rapid sequential ops, tight test loops).

Changes

• SyncRepository: Added int _logCounter = 0; per-instance field and updated _logId() to append it, matching the SyncManager._generateOpId() pattern.

// Before
String _logId() {
  final micros = DateTime.now().microsecondsSinceEpoch;
  return 'sync_log_$micros';
}

// After
String _logId() {
  _logCounter += 1;
  final micros = DateTime.now().microsecondsSinceEpoch;
  return 'sync_log_${micros}_$_logCounter';
}

Log IDs are now sync_log_<microseconds>_<counter>, guaranteed unique within a SyncRepository instance.

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[Copilot]

Pull request overview

Fixes potential SyncRepository._logId() collisions by adding a per-instance counter to the generated log IDs, aligning the scheme with SyncManager’s timestamp+counter pattern.

Changes:

• Add per-instance _logCounter state to SyncRepository.
• Update _logId() to generate sync_log_<micros>_<counter>.

---

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

[Copilot]

_logId() now guarantees uniqueness only within a single SyncRepository instance. If multiple repositories share the same SyncLogStore (or if a store persists across app restarts and the counter resets), collisions can still happen because the ID lacks a repository/device/component discriminator. Consider including a configurable prefix (e.g., repository/device ID) or passing a logIdNamespace/instanceId into the repository so IDs are unique across all writers to the same store.

[Copilot]

This change addresses a timing-resolution collision, but there’s no deterministic test covering it (calling add() in a tight loop may not reliably reproduce the original bug on all platforms). To make this behavior testable and prevent regressions, consider injecting a time provider/clock into SyncRepository (used by _logId() and SyncLog.timestamp) so a test can force identical timestamps and assert IDs remain unique.