=fix(cli): update in-progress request count immediately when requests are dispatched (#550)

[shobhitagnihotri69]

Summary

Fixes #550.

Previously, when requests were dispatched in process_requests_online, status_tracker.num_tasks_started and status_tracker.num_tasks_in_progress were incremented, but status_tracker.update_display() was never called. Furthermore:

1. Permanent failure blocks decremented in_progress and incremented failed without triggering an update_display().
2. In OnlineStatusTracker, state-change updates in TQDM/CLI mode were gated behind current_time - self._last_stats_update >= _STATUS_UPDATE_INTERVAL and (...), preventing state changes from reflecting until 5 seconds elapsed.

This caused the CLI to display In Progress: 0 while requests were in-flight to LLM providers.

Changes

• Call status_tracker.update_display() immediately after incrementing num_tasks_started and num_tasks_in_progress in process_requests_online.
• Call status_tracker.update_display() when requests fail permanently in handle_single_request_with_retries.
• Update OnlineStatusTracker.update_display() so task state transitions trigger console updates without delay.
• Add regression test test_online_status_tracker_requests_in_progress_update in tests/unittests/test_online_status_tracker.py.

Verification

• Ran pytest suite: 10/10 passed in tests/unittests/test_online_status_tracker.py and tests/unittests/test_status_tracker.py.
• Ruff check: clean (0 errors).

[entelligence-ai-pr-reviews]

EntelligenceAI PR Summary

Updates online request processing to refresh status displays immediately when requests start or permanently fail. It also allows state-change updates to bypass the periodic refresh interval and adds coverage for in-progress transitions and display formatting.

---

Review Scorecard

Dimension	Rating	Basis
Code Quality	●●●●○ 4/5 — Good	no critical/significant findings — rated Good
Blast Radius	Low	changed symbols are referenced only within their own file(s); no high-impact surface touched, 2 file(s) / ~4 line(s) changed (size only — not a blast signal)
Merge Confidence	●●●●○ 4/5 — Mostly Safe	code quality 4/5 × Low blast radius

Safe to merge with a low blast radius; code quality rated Good (4/5): the 1 comment are minor or cosmetic, with no runtime, logic or security impact.

Key Findings:

• No critical or significant issues found

Need to merge before these are addressed? Anyone with write access can comment @entelligence /approve to approve it now. @entelligence help lists every command.

Evaluated against

• 4/4 changed files reviewed
• criteria: correctness, security & access control, robustness & error handling, concurrency & data integrity, repo conventions / steering docs
• steering docs: none found in repo

[entelligence-ai-pr-reviews]

Walkthrough

This PR synchronizes online status displays with request lifecycle changes by refreshing progress immediately when requests start or fail after retries. It also updates display-trigger logic to react to task metric changes and improves test coverage and rendering compatibility for ellipsis characters.

Changes

Files

• src/bespokelabs/curator/request_processor/online/base_online_request_processor.py
• src/bespokelabs/curator/status_tracker/online_status_tracker.py

Updated online request processing to refresh the status display when requests enter progress and when exhausted retries produce generic responses and failure counter updates. Modified update_display() to refresh on task metric changes in addition to the time interval, including changes to in-progress, succeeded, and failed counts.

Files

• tests/unittests/test_online_status_tracker.py
• tests/unittests/test_status_tracker.py

Updated display assertions to accept either the three-dot leader (⋯) or Unicode ellipsis (…). Added lifecycle coverage confirming that dispatching and completing requests updates in-progress, started, succeeded, and maximum-concurrency counters.

[entelligence-ai-pr-reviews]

Exercise the TQDM path in this regression test

With the default USE_RICH_DISPLAY, start_tracker(console) selects the Rich branch, so this test never exercises the changed TQDM condition. It only checks counters and max concurrency, so it passes if the five-second TQDM gating remains broken.

Prompt to fix with AI

Copy this prompt into your AI coding assistant to fix this issue.

In tests/unittests/test_online_status_tracker.py at lines 127-148, make the regression test explicitly exercise the non-Rich/TQDM mode or otherwise isolate the TQDM update logic, and assert that the rendered/logged status changes immediately after incrementing and decrementing in-progress requests. Preserve the existing Rich-mode coverage separately if needed.