MBombeck/HealthLog v1.4.47.1

v1.4.47.1 — slim summaries 9s → ~0.5-1s

on GitHub

Same-day hotfix on v1.4.47. Dashboard cold mount was paying ~9 s TTFB on /api/analytics?slice=summaries for power-user accounts; tracked to a single SQL pass that scanned the user's full measurements partition without an outer time cap, even though every aggregate inside the SELECT restricted to 7/30/90 days.

Changed

• src/lib/analytics/summaries-slice.ts — the narrows query inside computeFromRollups gains AND m."measured_at" >= NOW() - INTERVAL '90 days' on the outer WHERE. The planner now uses an index range scan on (user_id, type, measured_at) instead of a full per-user partition read. Output is bit-identical because the inner FILTER (WHERE m."measured_at" >= NOW() - INTERVAL 'N days') clauses already discarded rows outside the 90-day window.

Effect

On a ~450 000-row tenant:

• slim slice cold: ~9 s → ~0.5-1 s
• slim slice warm: sub-50 ms (unchanged, Map-lookup cache)

The computeFromLiveAggregate split (for tenants without rollup coverage) stays queued for v1.4.48 — it needs a 2-query split to preserve the all-time count/min/max/mean.

Risk

Zero. No schema change, no migration, no env-var change. Standard image roll.

Operator notes

No prisma migrate deploy step required.