Skip to content

Dashboard: multi-policy samples, denser UX, chart polish - #75

Merged
leo-gan merged 8 commits into
masterfrom
fix/dashboard-1
Jul 29, 2026
Merged

Dashboard: multi-policy samples, denser UX, chart polish#75
leo-gan merged 8 commits into
masterfrom
fix/dashboard-1

Conversation

@leo-gan

@leo-gan leo-gan commented Jul 29, 2026

Copy link
Copy Markdown
Owner

Summary

  • Multi-policy sample sets for outlier research (all, iqr_1.5, iqr_3, winsorize_5_95) with schema 2.2 compact export (identity + variants) published as stats_*.json.gz
  • Dashboard Samples control rebinds tables, charts, and Pareto; filter provenance banner
  • Compare lab polish, denser page top, data type wording (not fixture)
  • Chart fixes: latency axis unit matches ticks; ranking legend Size then Latency; SORT is Size · Ops/s|Latency

Validation

  • Tests: analysis (84), python (29), javascript (9), rust (13) — pass
  • Benchmarks: skipped (no benchmark-runner language source changes vs master)
  • Error CSVs: n/a (no re-bench)
  • Analysis: n/a re-run (existing multi-policy stats packs already on branch)
  • Dashboard: sync-data.py verified; all 9 langs have *_latest.json.gz + stats_*_latest.json.gz (schema 2.2)

Notes

  • Full re-bench not required for this UI/analysis-export branch
  • Gzip-only recompression of identical dashboard blobs was not re-committed after the final gate sync

leo-gan added 8 commits July 28, 2026 11:02
Implement the dashboard improvement plan on fix/dashboard-1:

Phase 0–1: mobile hamburger nav; semantic badges (Optimal/Dominated, no
purple-for-negative); KPI loading/empty states; format.js (3-sig continuous
metrics, locale sizes, units in headers only, numeric alignment); merge
Side-by-Side + All Metrics + Cross-lang into one Compare lab with baseline
column and value (ratio×) cells; relative logs paths and honest History on
static hosting.

Phase 2–3: dual throughput/latency + size ranking chart; log X toggle on
scatter; run provenance strip; URL query state; lazy cross-lang load; Chart
PNG + Markdown export; slim sync payloads (omit CSV by default); Vite
cleans stale docs/dashboard/assets on build.
Refine the analytics UI after the Compare lab merge: denser Compare
toolbar with same-language Test Data/Mode/Language parity, fixed layout
bugs (hidden filters, stacked XL add controls), and clearer status copy.

Detailed Analytics now has its own Ops/s|Latency switch with Median/Std/
P95/P99 columns, Pareto via row highlight (no Frontier column), and
×base vs ×med ratio rules. Ranking uses a single diverging chart with
speed/size sort. Fixture dropdown adds batch compounds plus all@1,
all@100, and all@all. Run meta shows run id, host RAM, and labeled
runtime instead of git commit.
Export four named aggregations (all, iqr_1.5, iqr_3, winsorize_5_95) with
filter provenance, and let the dashboard switch sample sets for tables,
charts, and Pareto. Refresh published stats and language packs for all
languages.
Publish identity+variants packs (no duplicated groups), catalog-only
filter labels, and stats_*.json.gz instead of plain multi-megabyte JSON.
Replace user-visible "fixture" with "data type" in samples meta, KPIs,
notifications, and export notes.
Tighter header, toolbar, samples banner, run meta, and KPI spacing with
no copy changes.
Complete page-top density pass.
Align latency axis title with tick scale; put Size first in ranking
legend and SORT controls; label sort as Ops/s or Latency by view mode.
@gemini-code-assist

Copy link
Copy Markdown
Contributor

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

@leo-gan
leo-gan merged commit 079655c into master Jul 29, 2026
11 checks passed
@leo-gan
leo-gan deleted the fix/dashboard-1 branch July 29, 2026 00:26
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant