Metadata
| Status | done |
|---|---|
| Assigned | agent-745 |
| Agent identity | f51439356729d112a6c404803d88015d5b44832c6c584c62b96732b63c2b0c7e |
| Created | 2026-04-27T16:20:42.025213243+00:00 |
| Started | 2026-04-27T16:21:45.838412749+00:00 |
| Completed | 2026-04-27T16:52:40.466022492+00:00 |
| Tags | tui,bug, eval-scheduled |
Description
Description
The TUI HUD shows '1/8 tasks active' but the graph viz shows ~6 tasks actually in-progress. The HUD counter is undercounting.
Investigate:
- Where the HUD's 'N/M active' count is computed (likely src/tui/, search for 'active' or counter formatting)
- Where the viz determines a task is 'running' (highlighting, colour, etc.)
- Reconcile: are they using different filters? (e.g. HUD might count only one status like 'in-progress', viz might also include 'pending-validation' or agent-spawned states)
- Pick the right definition of 'active' (probably: status in {in-progress, pending-validation} + has live agent) and use it consistently in both places — extract a helper if not already present
Reproduction:
- User observed 1/8 in HUD vs ~6 highlighted as running in 'wg viz' / TUI graph view
- Likely reproducible right now on this graph since the chat-redesign batch is dispatching multiple agents
Validation
- Failing test first: a graph with K in-progress tasks reports 'K/total' in the HUD (not less)
- HUD count == count of tasks visually marked 'running' in the TUI viz, on the same snapshot
- Definition of 'active' is documented/inline-commented at the helper site
- Live-smoke: launch TUI on a graph with multiple in-progress tasks; HUD count matches highlighted-node count
- cargo build + cargo test pass
Depends on
Required by
- (none)
Log
- 2026-04-27T16:20:41.993175219+00:00 Task paused
- 2026-04-27T16:21:14.953456502+00:00 Task published
- 2026-04-27T16:21:45.144332567+00:00 Lightweight assignment: agent=Careful Programmer (f5143935), exec_mode=full, context_scope=task, reason=Only programmer available with proven track record (0.72 score on 111 tasks); Careful tradeoff aligns with bug-fix correctness requirements; task needs code reading and implementation.
- 2026-04-27T16:21:45.838417327+00:00 Spawned by coordinator --executor claude --model opus
- 2026-04-27T16:21:53.270415604+00:00 Starting investigation of HUD vs viz active-task count mismatch
- 2026-04-27T16:27:21.600813032+00:00 Identified mismatch: viz colors {InProgress, Waiting, PendingValidation} as yellow (running-like), but HUD task_counts.in_progress counts only {InProgress, PendingEval}. Extract Status::is_active() helper, unify.
- 2026-04-27T16:52:08.397514636+00:00 Validated: cargo build OK; new tests pass (Status::is_active + task_counts_active_tests). One pre-existing test failure (provenance_full_lifecycle_all_ops_recorded) unrelated to this change — confirmed by stashing edits and reproducing on clean tree.
- 2026-04-27T16:52:25.119509011+00:00 Committed: df447e4ca — pushed to remote
- 2026-04-27T16:52:40.466026259+00:00 Task pending eval (agent reported done; awaiting `.evaluate-*` to score)
- 2026-04-27T16:54:55.981830493+00:00 PendingEval → Done (evaluator passed; downstream unblocks)