v0.33.6

docs/TEST_REPORT.md

@@ -0,0 +1,44 @@
+# RustProxy 0.33.6 - verification report
+
+## Scope (2026-10-15)
+
+Additional 0.33.6 chart/UI changes:
+- All 14 full-width Overview and Web Server plots retain a fixed 125px SVG
+  height, regardless of viewport width and dashboard density.
+- Grid lines are mapped to the actual plotting range rather than approximate
+  horizontal positions.
+- Transform and Compression activity (plus other unbounded unit series) use
+  sane zero-based minimum ranges and monotonic, rounded range expansion, so
+  low values do not look like maximum utilization and spikes remain visible.
+- Full Cache and Transform live charts keep an 180px plot height and the same
+  stable per-second range policy. Compact percent sparklines stay 0-100%.
+
+Previous 0.33.5 regression scope:
+
+Correct the Overview **System and process CPU** widget:
+- Use a fixed 0-100% vertical range, not the measured-history peak.
+- Map 25%, 50% and 75% to the horizontal grid lines.
+- Clip plotted CPU values above 100% at the top without modifying their measured values in the legend or summary.
+- Fix only this plot's CSS height at 125px, independent of viewport width, dashboard density and CPU samples.
+- Leave other chart scales, plot dimensions, status sampling and dashboard preferences unchanged.
+
+## Checks completed
+
+- **PASS:** `python scripts/tests/dashboard_stable_scales_0336_browser_test.py`
+  with real bundled HTML/CSS/JS in simulated Chromium at 320, 375, 768,
+  1440 and 1920px and all three density modes. Covered 14 large plots,
+  125/180px plot heights, quarter-grid alignment, 0-100% CPU mapping,
+  steady low/high traffic plotting, minimum count/rate/latency/memory scales,
+  sticky expanded scales, reset and non-clipped live operation-rate labels.
+
+- **PASS:** `python scripts/tests/dashboard_cpu_fixed_scale_0335_browser_test.py` against the actual bundled HTML/CSS/JS in simulated Chromium, at 320, 375, 768, 1440 and 1920px, in normal/compact/ultra densities. Checked low and high CPU histories, missing samples, new status revisions, clipping above 100%, actual percentage labels, consistent grid positions, fixed plot height and no sample-dependent changes to the CPU/adjacent Connection Activity cards' heights.
+- **PASS:** All 11 prior `scripts/tests/*browser_test.py` simulated-browser regressions (including the 0.33.5 CPU test) and the new 0.33.6 browser regression: chart headers, resized widths, dashboard grid/modal resizing, sidebar, progress bars, System RAM, Web observability and Web Server metrics. No live RustProxy server was used.
+- **PASS:** `node scripts/tests/overview_samples_0330_ui_test.mjs`, `node scripts/tests/dashboard_resume_rate_charts_ui_test.mjs` and `node scripts/tests/dashboard_metric_visuals_ui_test.mjs`.
+- **PASS:** `./dev.sh doctor` and `python scripts/preflight.py` (source layout). 
+- **PASS:** All eight `migrations/*.sql` applied successfully in two successive cycles to a disposable, in-memory SQLite database. This verifies SQL script execution/idempotency in this context, not an upgrade of production data.
+- **PASS:** Parsed `scripts/deploy/upgrade.py --help` and `scripts/migration/migrate_npm.py --help`; did not run a live upgrade or migration against an external database or service.
+
+## Not verified in this environment
+
+- `./dev.sh check` and `./dev.sh test` both exit 127: Cargo/rustc are unavailable. Therefore no Rust compilation, unit/integration suite, actual proxy runtime, real SSE feed, production database upgrade or live load test has been verified.
+- The real-runtime acceptance gate remains open until Rust builds and runs against representative test traffic.