Conversation
There was a problem hiding this comment.
💡 Codex Review
apply_bespoke_event_handling no longer takes an analytics_events_client parameter, but this block still reads analytics_events_client.as_ref(). That leaves an unresolved identifier and breaks compilation of codex-app-server before runtime tests can run. The same stale reference pattern appears again later in this function.
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| started_at_ms: None, | ||
| completed_at_ms: None, |
There was a problem hiding this comment.
Populate command approval completion timestamps
complete_command_execution_item emits ItemCompleted with started_at_ms/completed_at_ms set to None. The new analytics reducer drops command tool events when either timestamp is missing (option_i64_to_u64(...)?), so denied/timed-out guardian command executions never produce codex_command_execution_event even though their statuses are mapped.
Useful? React with 👍 / 👎.
| started_at_ms: None, | ||
| completed_at_ms: None, |
There was a problem hiding this comment.
Preserve file-change timing on denied approvals
Declined/failed file-change approvals are emitted as ItemCompleted with all timing fields None. In the analytics reducer, file-change events require non-null started_at_ms and completed_at_ms, so these rejected outcomes are silently dropped and no codex_file_change_event is recorded.
Useful? React with 👍 / 👎.
|
Closing duplicate stack created during resubmission; canonical PR remains #18747. |
Stack created with Sapling. Best reviewed with ReviewStack.