Slate v2 clipboard proof must split fragment semantics from DOM transport

What happened

While landing the slate-v2 clipboard-boundary proof, the first few attempts kept failing architect review for the same reason:

• the plan or code treated clipboard as mostly a DOM transport problem
• but the real missing seam was partly in slate-v2 itself

The proof only became stable after the boundary was split explicitly:

• slate-v2 owns fragment meaning:
  • Editor.getFragment(editor)
  • Transforms.insertFragment(editor, fragment, options?)
• slate-dom-v2 owns browser transport:
  • MIME keys
  • DataTransfer
  • HTML scraping
  • plain-text export/fallback
  • payload provenance/version envelope checks

Why this mattered

If DOM transport owns fragment meaning, the browser layer starts deciding what the editor document is.

If core owns MIME/HTML transport, the engine starts swallowing browser junk.

That blur already existed in legacy Slate across:

• core fragment extraction
• DOM clipboard write/read helpers
• React event wiring
• downstream getFragment / insertFragment overrides

The proof stayed honest only after that blur was made explicit and then reduced.

What fixed it

1. Pin one narrow core seam first.
2. Keep DOM transport outside core.
3. Use a versioned clipboard namespace for the v2 proof instead of reusing the legacy one.
4. Make unsupported payloads and unsupported copy/paste targets fail closed instead of throwing.
5. Keep the proof subset narrow and explicit: top-level blocks with one text child, ordinary text/block paste targets.

Reusable rule

For editor clipboard architecture:

• core decides fragment meaning
• DOM decides browser transport
• React only delegates

If a clipboard proof cannot state those three lines cleanly, it is not done.