Maintenance Mainline

MoUI keeps a narrow default maintenance baseline so the framework can keep growing without turning every diagnostic route into daily release pressure.

Tier	Canonical routes	Gate
Tier 1	macOS Skia, Web wasm-gc WebGPU (Canvas2D fallback)	Blocking: PR build/test, daily presentation, and release evidence
Tier 2	Windows Skia, Linux Skia	Blocking: L0-L2 and first frame; complete L3 may remain partial
Tier 3	macOS WGPU (CoreText, Cosmic fallback), Windows WGPU (DirectWrite, Cosmic fallback), Linux WGPU (Fontconfig, Cosmic fallback), Android window-hosted Skia, iOS window-hosted Skia, HarmonyOS window-hosted Skia, macOS Sun, Windows Sun, Linux Sun, WeChat Skyline Canvas2D	Non-blocking: scheduled/manual build, run, and evidence

Tier, L0-L3 evidence, and product_class/ready are independent. Source: checks/platform-matrix.json; actual observations remain in checks/platforms/*.json.

Status Classes

• mainline: covered by the default sh scripts/check.sh --profile daily path. Mainline work must stay green before ordinary handoff.
• diagnostic: runnable, testable, and allowed to record observation, but not a default daily gate. Run the matching opt-in flag or focused command when the diagnostic route changes.
• pending: documented scaffold or capability note exists, but the route must not be described as ready until matching smoke has run.

Default Baseline

The default daily baseline covers:

• moui/core and moui/views
• moui/backend and moui/backend/web
• moui/render, moui_skia_renderer, and moui_web_renderer
• fallback-safe moui_skia checks
• Showcase app/Web wasm-gc validation
• renderer/provider static checks and lightweight Web handoff validation
• maintenance baseline ratchets for oversized source files, source-level pub(all) counts, and root facade type-forwarding counts
• API surface package budgets plus semantic classification budgets for core, views, runtime, backend, and render

Complete platform runtime smoke is a release or matching-host gate, not a default daily gate. Daily checks do not require fresh matching-host promotion.

Engineering Baseline Ratchets

The maintenance baseline ratchets are implemented as the MoonBit tool at tools/moui/validate_maintenance_baseline. They scan MoUI-owned MoonBit source under moui/, examples/, and website/, excluding generated pkg.generated.mbti files, vendored .mooncakes/ trees, build output, and generated Unicode fixture tests. The default sh scripts/check.sh --profile daily path runs the guard through node scripts/validate-maintenance-baseline.mjs; run that wrapper directly when splitting files, shrinking pub(all), or reducing root facade forwards, then ratchet the relevant budget downward in the same change.

The guard tracks three budgets:

• oversized file ratchets for current hotspots such as moui/core/unicode/grapheme_data.mbt, moui/runtime/view_tree.mbt, moui/views/form/form_validation_test.mbt, moui/backend/common lifecycle/image tests, Skia renderer text/tests, and large example app files;
• direct package source pub(all) counts for core, views, host, render, mainline examples, Mo Workbench, PDF Workbench, and Website;
• root facade pub type forwarding count in moui/moui.mbt.

The API surface guard is separate from this maintenance baseline. It tracks generated public API size, forbidden boundary tokens, and semantic API classification budgets such as app_constructor, advanced_core_protocol, runtime_diagnostic, host_contract, and renderer_contract. See API surface and API surface audit before expanding public API.

Application performance guardrails are also separate from source-size and API ratchets. checks/performance-budgets.json owns the reviewed thresholds and update policy; node scripts/validate-performance-budgets.mjs validates the catalog in the PR profile, while the macOS benchmark job compares real Skia Raster samples and uploads the ignored result artifact. Do not weaken a threshold to accommodate an unexplained regression or compare artifacts from a different runner.

Current max values intentionally match today's debt so unrelated changes do not need to solve the whole backlog. When a refactor splits a file, moves control-level entrypoints into views, shrinks public surface area, or removes root facade forwards, lower the corresponding budget in the same commit.

The default wrapper runs the daily scope. The MoonBit tool also supports a full-workspace hotspot scope:

moon run tools/moui/validate_maintenance_baseline --target native -- --scope full

The full scope keeps the daily budgets and additionally scans addon/tool workspace roots such as moui_richtext, moui_skia, moui_sun, moui_theme, moui_tests, moui_devtools, moui_webview, moui_agent*, and tools. Known large files have explicit full-only temporary budgets so broad validation can report hotspots without forcing every 2k-4k line diagnostic/test file into the daily baseline or this refactor. Split those files in focused follow-up changes and ratchet their full-only budgets down as each split lands.

Diagnostic Routes

Native WGPU is an experimental renderer diagnostic route. Its engineering gate stays diagnostic — runnable, testable, allowed to record observation, but not a default daily gate and carrying no product commitment. It stays available through sh scripts/check.sh --profile full and focused renderer/provider commands, but it is not the native mainline.

Design Systems is addon diagnostic coverage. moui_theme and examples/design_systems remain important source-mapped preview/parity surfaces, but they are not part of the core MoUI framework baseline. Run sh scripts/check.sh --profile theme when changing moui_theme or the Design Systems example.

Slow native example builds and matching-host platform runtime collection remain opt-in through --profile full, real Skia smoke helpers, and direct runtime logs.