daily-fix: 1-cell pilot timing for per-cell fit phases + 2x
kind: infra#wf-fix#wf-fix-fp:04e62ed208f3#daily-auto-filed
track:Highlight text on any card (body, plan, or an event) to anchor a comment, or leave a whole-task comment here. Mention @claude to summon a reply.
No comments yet.
Overview / Motivation
Auto-filed by /daily 2026-07-04 (route 2: behavior/logic change -> independent review), from the nightly transcript problem sweep.
Goal
plan-compute-sizing requires a MEASURED 1-cell pilot timing x cell count for any per-cell ridge/SVD/GD fit phase before accepting the section-9 estimate; and the vectorize rule's Supersede contract gains a mid-run trigger: a compute-deviation at 2x the plan estimate forces the vectorize check immediately.
Workflow gap
- Bug observed: #811's per-cell ridge-fit phase ran ~19h against a ~6h plan estimate on 2026-07-04 (42/108 cells at 19h21m, ~4x projection) and #931's p3_fits overran ~2x - the recurring #823-class serial dense-fit sizing failure; the #811 session only pivoted to vectorizing after flagging a second compute-deviation, ~19 CPU-h in.
- Why it is a workflow gap: the failure originates in the workflow surface / helper named below, not in any one experiment.
- Confidence (emitter): medium
Scope / surfaces
- Primary target:
.claude/rules/plan-compute-sizing.md, .claude/rules/vectorize-many-cell-fits.md - Sessions: 86923fc5 + f2ef929f (#811), 2f37cc81 (#931).
Constraints / invariants
- Workflow-surface only;
scripts/workflow_lint.py --check-asks+--check-referencesstay green; ruff clean on touched files.
Provenance
- workflow_fix_target: .claude/rules/plan-compute-sizing.md, .claude/rules/vectorize-many-cell-fits.md
- source: /daily 2026-07-04 problem sweep (transcript-mined)
Activity