perf(planning): restore the integrated benchmark gate #278

Merged

nessar merged 5 commits from feature/benchmark-performance into develop 2026-10-16 22:36:32 +02:00

Conversation 0 Commits 5 Files Changed 15

+853 -125

nessar commented 2026-10-16 08:43:45 +02:00

Owner

Copy link

Copy source

The integrated benchmark previously exceeded its unchanged 300-second timeout. Reduce candidate-scoring and allocation work so the complete seven-scenario protocol finishes in 298.551 seconds locally, then restore benchmark success as a requirement for image publication and the aggregate CI check.

Evaluate objectives lazily in priority order, reuse calendar geometry and working-day prefixes, and share existing placements during assignments while detaching mutable metadata. Preserve search breadth, rule order, earned-entitlement chronology, outcomes, explanations and deterministic tie-breaks. Benchmark inputs, assertions, warm-ups, repetitions and resource limits remain unchanged.

Validation:

• Required benchmark: exit 0 with all seven warm-ups and 14 measured runs; 2048 MiB heap, no coverage, 300-second timeout.
• 633 functional tests with coverage; application/spec type checks, lint, formatting, CI policy/image tests and production build pass.
• Complete baseline comparisons match for all seven scenarios and 182 configuration cases; three permanent regression cases pin objective vectors and provenance.
• Native-worker browser completion, input editing, heartbeats and keyboard cancellation checks pass. Independent Standards and Spec reviews found no findings.

Evidence and reproduction: docs/benchmark/issue-78.md, including all 14 protocol rows. The local timeout margin is only 1.449 seconds; live CI and security scans are unverified. No physical-phone run was available, and calendar rendering can pause the main thread for about two seconds.

Refs #78. The related verification issue #77 / PR #273 remains open; this PR does not close either issue or change the #268 activation gate.

The integrated benchmark previously exceeded its unchanged 300-second timeout. Reduce candidate-scoring and allocation work so the complete seven-scenario protocol finishes in 298.551 seconds locally, then restore benchmark success as a requirement for image publication and the aggregate CI check. Evaluate objectives lazily in priority order, reuse calendar geometry and working-day prefixes, and share existing placements during assignments while detaching mutable metadata. Preserve search breadth, rule order, earned-entitlement chronology, outcomes, explanations and deterministic tie-breaks. Benchmark inputs, assertions, warm-ups, repetitions and resource limits remain unchanged. Validation: - Required benchmark: exit 0 with all seven warm-ups and 14 measured runs; 2048 MiB heap, no coverage, 300-second timeout. - 633 functional tests with coverage; application/spec type checks, lint, formatting, CI policy/image tests and production build pass. - Complete baseline comparisons match for all seven scenarios and 182 configuration cases; three permanent regression cases pin objective vectors and provenance. - Native-worker browser completion, input editing, heartbeats and keyboard cancellation checks pass. Independent Standards and Spec reviews found no findings. Evidence and reproduction: [docs/benchmark/issue-78.md](https://gitea.nessar.fr/nessar/vac-optimizer/src/commit/859a5f11052a8e2d71321b1a685177b7f44588b3/docs/benchmark/issue-78.md), including all 14 protocol rows. The local timeout margin is only 1.449 seconds; live CI and security scans are unverified. No physical-phone run was available, and calendar rendering can pause the main thread for about two seconds. Refs #78. The related verification issue #77 / PR #273 remains open; this PR does not close either issue or change the #268 activation gate.

nessar added 1 commit 2026-10-16 14:01:48 +02:00

perf(planning): restore the integrated benchmark gate ... cbde23f572

Reduce objective evaluation and calendar allocation work while preserving planner outcomes and deterministic explanations. Restore required benchmark gating and record the unchanged protocol, differential checks, and browser evidence for #78.

nessar force-pushed feature/benchmark-performance from 859a5f1105 to cbde23f572 2026-10-16 14:01:48 +02:00 Compare

nessar referenced this issue from a commit 2026-10-16 22:24:23 +02:00

docs(benchmark): record PR #278 benchmark headroom

nessar added 4 commits 2026-10-16 22:24:23 +02:00

perf(planning): build placement date keys directly c7e7abcd24

perf(planning): reuse immutable non-working date counts 433a8b0ade

perf(planning): reject impossible mandatory moves before copying ... c6610cd633

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

docs(benchmark): record PR #278 benchmark headroom ... 1711bc570c

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

nessar referenced this pull request 2026-10-16 22:34:15 +02:00

perf(planning): incremental exact scoring to reach a 10-second loaded year #279

nessar added a new dependency 2026-10-16 22:34:22 +02:00

#279 perf(planning): incremental exact scoring to reach a 10-second loaded year

nessar merged commit a371e03462 into develop 2026-10-16 22:36:32 +02:00

nessar referenced this pull request 2026-10-16 22:39:45 +02:00

test(planning): compare the engine against a frozen engine on random requests #280

nessar added a new dependency 2026-10-16 22:39:45 +02:00

#280 test(planning): compare the engine against a frozen engine on random requests

nessar referenced this pull request 2026-10-16 22:39:45 +02:00

test(benchmark): report each scenario as it finishes #281

nessar added a new dependency 2026-10-16 22:39:45 +02:00

#281 test(benchmark): report each scenario as it finishes

nessar referenced this pull request 2026-10-16 22:39:46 +02:00

ci(benchmark): gate scenario speed relative to the runner #285

nessar referenced this issue from a commit 2026-10-17 00:52:35 +02:00

test(planning): compare the engine against a frozen engine on random requests

Sign in to join this conversation.

Reviewers

No Reviewers

Labels

Clear labels

bug

dependencies

docker

documentation

duplicate

enhancement

github_actions

good first issue

help wanted

invalid

javascript

needs-info

needs-triage

question

ready-for-agent

wontfix

No labels

Milestone

No items

No Milestone

Projects

Clear projects

No projects

Assignees

Clear assignees

renovate (renovate) Bot nessar (nessar)

No Assignees

1 Participants

Notifications

Subscribe

Due Date

No due date set.

Blocks

#279 perf(planning): incremental exact scoring to reach a 10-second loaded year

nessar/vac-optimizer

#280 test(planning): compare the engine against a frozen engine on random requests

nessar/vac-optimizer

#281 test(benchmark): report each scenario as it finishes

nessar/vac-optimizer

Reference: nessar/vac-optimizer#278