feat(bench): B2 follow-up #5b — sort + filter comparator wiring#131
Merged
Conversation
Plan for the second slice of B2 follow-up #5: open the sort / filter-metadata / filter-text gate for ag-grid + tanstack + mui, wire library-native sort + filter dispatch in each adapter, re-run the matrix to capture comparator data for H6/H7/H8. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Drop the pretable-only gate on sort/filter-metadata/filter-text — all four bench adapters now wire their native sort/filter API. The scenario gate (S2/S7 only) remains. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…lumnState + setFilterModel Wire the interactionPlan prop into the AG Grid adapter so the bench harness can drive sort/filter via the library's native gridApi. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ting + setColumnFilters Capture the Table instance in a ref synced each render so a useEffect can drive sort/filter via TanStack's library-native API. filter-metadata mode uses the equalsString filterFn to match the bench plan's equals semantics. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ortModel + setFilterModel Wire the interactionPlan prop into the MUI adapter so the bench harness can drive sort/filter via the library-native apiRef. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Drop the pretable-only gate around measureBenchInteractionRun. All four adapters now run sort + filter through the interaction probe. Telemetry-based state reading remains pretable-only; comparators fall back to DOM-default state reading (undefined override). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…idence Append the 2026-05-10 entry covering the gate-opening + per-adapter wiring, n=3 latency table, and H6/H7/H8 status delta. Ship the matrix re-run milestone alongside. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Contributor
Vercel preview readyPreview: https://pretable-k6e1ayb7q-cacheplane.vercel.app Updated automatically by the |
blove
added a commit
that referenced
this pull request
May 11, 2026
Three editorial surfaces (ComparisonTable interaction rows + trail-marker labels + new /bench Interactions section) driven by the PR #131 sort + filter comparator data. No source changes; new aggregated milestone summary file built from per-run JSONs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
4 tasks
blove
added a commit
that referenced
this pull request
May 11, 2026
* docs(plans): B2 follow-up — homepage interaction wedge refresh plan Six-task plan: aggregator script + milestone summary, ComparisonTable rows + labels, tests, /bench Interactions section, repo-memory, gates + PR (no auto-merge). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore(bench): aggregator for sort/filter milestone summary * fix(website): ComparisonTable adds 3 interaction rows + trail-marker label refresh * test(website): update ComparisonTable trail-marker regex assertions Three labels changed in the prior commit (AG Grid + TanStack + MUI X) to reflect the interaction wedge from PR #131. Regex assertions updated to match the new label phrasings: - AG Grid: "1.7× slower scroll, 3× slower interaction; row-height drift" - TanStack: "Headless; ~2× slower interaction (filter-metadata ties pretable)" - MUI X: "Scroll-p95 parity; 2× slower interaction" The pretable "Recommended path" assertion is unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(website): /bench page Interactions section with sort + filter comparator data Replaces the placeholder paragraph that said "comparative interaction evidence is on the roadmap" with a real section paralleling the H1 scroll layout. Driven by the new aggregated milestone summary at status/milestones/2026-05-10-b2-sort-filter-summary.json (built from PR #131's per-run summaries). New loader (loadInteractionSummary) + verdict helper (interactionVerdictFor) mirror the existing scroll-side patterns. The verdict helper computes per-script ratios against the fastest adapter and annotates the TanStack filter-metadata tie inline ("2.1–2.6× slower (filter-metadata ties pretable)"). Per-adapter latency table + two prose paragraphs match the H1 section's structure for visual consistency. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * docs(research): repo-memory entry for homepage interaction wedge refresh Documents the three editorial surfaces touched (ComparisonTable rows + trail-markers, /bench Interactions section, aggregated summary file), the deliberate non-goals (ReceiptsBand owned by PR #129, n=20 follow-up deferred), and the remaining open threads from B2. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore: prettier-format milestone JSON + plan/spec docs Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
sort/filter-metadata/filter-textgate invalidateSupportedP0aRequestto all four adapters on S2/S7 (was pretable-only).BenchInteractionPlanvia its native library API:gridApi.applyColumnState+setFilterModeltable.setSorting+setColumnFilters(Table instance captured in a ref synced each render;filterFn: equalsStringforfilter-metadata)apiRef.current.setSortModel+setFilterModelbench-app.tsx's interaction dispatch to all four adapters; comparators passundefinedforreadInteractionStateOverride(telemetry override stays pretable-only). All four adapters now receiveinteractionPlan={interactionPlan}.S2 × {sort, filter-metadata, filter-text} × 4 adapters × 3 repeats(~2 min wall-clock). Milestone:status/milestones/2026-05-10-b2-sort-filter-comparators.hypotheses.json.Hypothesis status delta
H6/H7/H8 evaluators remain pretable-only — comparator data lives in the per-run
status/chromium-<adapter>-default-s2-hypothesis-{sort,filter-metadata,filter-text}-*.summary.jsonfiles.Per-adapter latency (n=3 medians;
interaction_latency_ms/settle_duration_ms)sortfilter-metadatafilter-textFindings: Pretable beats AG Grid Community 3–3.5× on every interaction script (16–18 ms vs 50–58 ms). Pretable beats MUI 2× across the board. TanStack mixed: filter-metadata latency narrowly beats pretable (15.7 ms vs 16.0 ms) but settle is 1.6× slower; sort/filter-text slower on both axes.
What's NOT in this PR
status/chromium-…summary.json; would require a new multi-adapter evidence shape).data-bench-result-row-countreads the full dataset; latency/settle metrics are still valid because the bench-runtime settle loop watches DOM signature change).Test plan
pnpm --filter @pretable-internal/bench-runner test— bench-runner test rewrites positive comparator parity assertionspnpm --filter @pretable/app-bench test— comparator interaction-dispatch test flipped to positive assertionpnpm -w typecheckpnpm -w lintpnpm formatpnpm bench:matrix … --scripts=sort,filter-metadata,filter-text— 36/36 runs completed across all four adapters🤖 Generated with Claude Code