Slate Browser Prioritized Backlog

Specialist testing/proof doc. For current queue and roadmap truth, see master-roadmap.md.

Purpose

This is the prioritized meaningful-test backlog for slate-browser.

It is derived from the proof work already completed in slate, slate-dom, and slate-react.

Strong rule:

• every item here should prove a real public behavior
• no “increase coverage” chores
• no implementation-detail snapshots pretending to be progress

Priority 0

These are the first tests that should exist next because they directly unblock the remaining architecture questions.

1. Empty-block IME composition on placeholder path

Lane:

• Playwright + Chromium CDP

Why first:

• this is the exact missing proof behind the zero-width renderer policy
• current legacy policy is still conservative because this path is unproved

What to assert:

• final committed text
• final Slate selection
• no crash or empty-state wedge

2. Browser contract fixture for DOM selection assertions

Lane:

• Vitest browser

Why first:

• the framework still lacks assertSelection / assertDomSelection
• many past proof slices already depend on selection being first-class

What to assert:

• one semantic selection snapshot
• one DOM selection snapshot
• exact mapping between them where appropriate

3. Clipboard contract fixture

Lane:

• Vitest browser for DOM-scale behavior
• Playwright for full example flow later

Why first:

• clipboard is already proved in v2 core/dom, but the framework API still lacks a clean clipboard contract lane

What to assert:

• DOM/plaintext/html contract behavior
• zero-width cleanup on copy

Priority 1

These build directly on the completed proof slices and raise the confidence bar without exploding scope.

4. Persistent annotation anchor example regression

Lane:

• Playwright

Why:

• RangeRef + persistent projections are already proved at runtime level
• now they need one mounted example regression

What to assert:

• anchor survives text insert
• anchor survives fragment insert
• rendered overlay remains attached to the right content

5. Zero-width shape contract fixtures

Lane:

• Vitest browser

Why:

• the bridge now supports both FEFF-shaped and -shaped line-break placeholders
• those shapes should stay pinned explicitly

What to assert:

• FEFF path
• path
• round-trip DOM/Slate offsets

6. Shared Playwright helper assertions

Lane:

• Playwright

Why:

• helper smoke exists, but assertions are still too thin

What to add:

• assertEditorHtml
• assertSelection
• assertDomSelection

Priority 2

These are important, but should wait until the first framework tranche is less rough.

7. More IME regressions

• Safari/WebKit composition ordering
• placeholder-visible first-character regressions
• delete-after-composition regressions

8. Agent-native lane

• only after deterministic lanes are stable
• only with explicit artifacts

9. Perf lane expansion

• rerender breadth
• huge-doc interactions
• placeholder/selection hot paths only when benchmark questions are clear

Mapping Back To Past Work

Completed proof work that should drive tests first:

1. clipboard boundary
2. projection store
3. range refs
4. persistent RangeRef-backed projections
5. zero-width DOM bridge normalization
6. placeholder bridge tolerance

Testing priority follows the remaining uncertainty around those proofs, not the chronological order they were implemented.