Slate DOM v2 Zero-Width Shape Strategy Proof Plan

Goal

Prove the next honest slice of the larger sentinel-strategy question:

• which zero-width DOM shapes the slate-dom-v2 bridge can already support
• where is viable
• where FEFF-like sentinel text still stays because of IME/native selection debt

Working Hypothesis

The first real proof should not try to remove sentinels globally.

It should:

• prove bridge compatibility for -style line-break placeholders
• keep current sentinel read/write normalization intact
• capture the policy line: line-break placeholders can move toward shape, but IME-sensitive empty inline/void/start-of-block cases still keep sentinel text for now

Phases

Phase 0

Ground the legacy DOM shape and current v2 bridge tolerance.

Phase 1

Add red tests for placeholder bridge behavior.

Phase 2

Implement only the missing bridge tolerance if needed.

Phase 3

Verify, deslop, re-verify, architect review, cleanup.

Progress Log

2026-04-03

• grounded the strategy seam against:
  • the approved zero-width round-trip bridge proof
  • legacy string.tsx zero-width rendering
  • legacy dom-editor.ts zero-width DOM translation
• chose the first honest cut: prove whether -style line-break placeholders already work through the slate-dom-v2 bridge before pretending to redesign the whole sentinel story
• added bridge tests for:
  • toDOMPoint over line-break placeholders
  • toDOMRange over line-break placeholders
  • toSlatePoint from both [br, 0] and [zeroWidth, 1]
• result: the bridge already supports -style line-break placeholders with no product-code changes required
• architect verdict: APPROVE
• deslop pass found no worthwhile simplification beyond the proof tests
• verification evidence:
  • zsh -lc 'yarn workspace slate-dom-v2 test'
  • zsh -lc 'yarn build:rollup'
  • zsh -lc 'yarn lint:typescript'
  • LSP diagnostics 0 on changed files
• reusable note captured in: 2026-04-03-br-line-break-placeholders-are-already-bridge-compatible.md