Skip to content

v0.5.4.1 — Mobile UX Hardening

Pre-release
Pre-release

Choose a tag to compare

@jaseknighter jaseknighter released this 31 Jan 05:10
· 2 commits to main since this release

Release Notes: v0.5.4.1 — Mobile UX Hardening & Social Serialization

Overview:
This maintenance release focuses on "Governance over Guesswork." Following the v0.5.4 Proof of Concept, v0.5.4.1 implements critical safety boundaries for the mathematical pipeline, resolves iOS Safari-specific UI regressions, and provides a transparent framework for experimental Meta-Testing diagnostics.


1. Safety & Mathematical Governance

  • Viewport "Explosion" Protection: Implemented a View-Safety Invariant in the math pipeline. Geometry now utilizes logarithmic dampening when vertex displacement exceeds 3x the base radius. This prevents exponential "explosions" during high pinch or spiral interactions while preserving creative expression.
  • Audit Tool Integrity: Fixed a runtime ReferenceError in the Formula Snapshot tool by correctly mapping Math.log within the execution scope.

2. Mobile UX & iOS Safari Hardening

  • Safari UI Alignment: Resolved a critical occlusion bug where the Micro-Nav toggle was hidden by the dynamic iOS Safari URL bar. The UI now respects env(safe-area-inset-bottom) to ensure visibility across all mobile browsers.

  • Gesture Conflict Resolution: Hardened the 3D canvas against browser-level interference.

  • Native pinch-to-zoom is now disabled on the canvas via touch-action: none.

  • Remapped Zoom: 3D zoom is now mapped to two-finger vertical swipes, providing a stable and predictable interaction model for mobile users.

  • HUD Drag Stability: Applied touch-action: none to the HUD header to prevent the browser from reclaiming drag gestures as scrolls, ensuring smooth dragging on mobile devices.

  • Tooltip Suppression: Finalized the block on MathTooltip for mobile layouts to prevent "sticky" hover artifacts typical of touch interfaces.


3. Meta-Testing Framework (Experimental)

  • Diagnostic Transparency: Updated the README to define the three-tier evaluation hierarchy used by the analyze-tests.cjs auditor:
    • Level 1 — Functional Tests: Standard Playwright/Jest execution (The primary CI Gate).
    • Level 2 — Meta-Tests: Evaluation of "Intent Visibility." The auditor scans test files for a match between a declared Semantic Tag and the presence of specific Code Markers (e.g., ensuring a [policy] tag is backed by ReadOnly or toThrow assertions).
    • Level 3 — Pipeline Self-Validation: A "Meta-Meta" layer where the auditor's own heuristic engine is tested against known-good and known-bad test samples to ensure report integrity.
  • Heuristic Definitions: Formally documented the regex-based markers for active tags used to bridge the Intent Visibility Gap:
    • [behavior]: Matches fireEvent, userEvent, click, postMessage.
    • [policy]: Matches toThrow, ReadOnly, frozen, Boundary, sanitize.
    • [failure-mode]: Matches error, NaN, Infinity, invalid.
  • Scoring Logic: Implementation of a "Signal Strength" gradient (🟢 Strong / 🟡 Weak / 🔴 Mismatch) based on the presence of a Tag, a corresponding Marker, and a non-trivial assertion (e.g., moving beyond toBeDefined).

4. Quality Audit (v0.5.4.1 Final)

Category Metric Result Notes
Global Pass Rate Total Success ✅ 100% (447/447) Unified Tier 1 & 2
Full Stack Statement Coverage ✅ 87.74% Unified Baseline
Visual Invariants Snapshot Integrity ✅ 100% Re-rolled for View-Safety
Meta-Audit (PoC) Intent Signal POC under evaluation

v0.5.4.1 | 2026-01-30 | Baseline Locked.