Slate React Deleted Test-Family Closure Consensus Plan

RALPLAN-DR Summary

Principles

1. restore contributor-facing proof, not dead internal machinery
2. keep file accounting exact: every deleted row ends as mirrored, restored, or explicit skip
3. do not bloat runtime.tsx into unreadable garbage when one focused contract file is cleaner
4. cut legacy internals directly when the current architecture no longer owns them

Decision Drivers

1. packages/slate-react/test/** is now the highest-value remaining deleted-test family
2. the deleted files mix real public behavior with obvious dead internals
3. runtime.tsx already proves a lot, but it should not become the only place every React claim goes to die

Viable Options

Option A: Add one self-contained focused contract file and keep runtime.tsx for the already-mirrored rows

Pros:

• keeps restored React-family proof readable
• preserves exact deleted-row accounting
• avoids fake legacy resurrection
• fastest path to execution without test-harness refactoring first

Cons:

• adds one more test artifact to maintain
• duplicates a small amount of mount/setup code unless extraction becomes worth it

Option B: Extract shared slate-react test helpers from runtime.tsx, then split restored proof into focused files

Pros:

• cleanest long-term test architecture
• avoids helper duplication between runtime.tsx and the new contract file

Cons:

• slower batch
• risks turning a deletion-closure task into a test-harness refactor

Option C: Widen packages/slate-react/test/runtime.tsx until it covers everything

Pros:

• no new test file
• reuses the current runtime harness

Cons:

• keeps stuffing unrelated proof into a giant file
• makes maintainer mapping harder, not easier

Option D: Restore the deleted upstream files as-is

Pros:

• easiest raw diff comparison
• lowest up-front judgment cost

Cons:

• reintroduces dead harness and dead architecture
• lies about the current package shape
• fights the roadmap rule against file-for-file resurrection

Chosen Option

• Option A

Why chosen:

• it gives one clean owner for the still-uncovered React-family value without pretending the old test tree still deserves to exist

Task Statement

Close the deleted packages/slate-react/test/** family by mapping each deleted file to current proof, restoring the still-relevant public-value gaps into one focused current contract, and explicitly cutting the dead internal residue.

Scope

In Scope

• the eight deleted packages/slate-react/test/** rows
• current proof ownership mapping into:
  • packages/slate-react/test/runtime.tsx
  • one new focused React-family contract file
• package code only if restored proof exposes a real slate-react behavior gap
• ledger and roadmap-note updates needed to close the family honestly
• one family closeout note

Out of Scope

• packages/slate-history/test/**
• supporting example/browser deletion families
• new runtime experiments outside the current Slate React contract
• resurrecting the deleted chunking implementation tree

Decision

Treat this family as three buckets:

1. mirrored now
   • deleted rows whose contributor-facing claim is already proved in packages/slate-react/test/runtime.tsx
2. restore now
   • deleted rows whose public-value claim still belongs but is not yet directly proved on the current surface
3. explicit skip
   • deleted rows tied to dead internal chunking or dead harness setup

Execution shape:

1. freeze the file-accounting table below before edits
2. keep packages/slate-react/test/runtime.tsx as the owner for already mirrored rows
3. add one new focused file:
   • /Users/zbeyens/git/slate-v2/packages/slate-react/test/surface-contract.tsx
   • this file does not exist yet and must be created, then assigned explicit proof ownership in the ledgers
4. use that file for the still-uncovered public-value rows:
   • ReactEditor.focus initializes selection from null
   • ReactEditor.focus does not spuriously fire onValueChange
   • structured render-surface mount stability
   • Editable translate attribute
   • useSelected plus structured path-rebasing stability
   • surface-contract.tsx starts self-contained; only extract shared helpers from runtime.tsx if a second restore case would otherwise duplicate the same mount/setup logic again
5. explicitly skip dead internal chunking and the deleted test-local tsconfig.json
6. create the family closeout note:
   • /Users/zbeyens/git/plate-2/docs/plans/2026-04-09-slate-v2-slate-react-deleted-test-family-closure.md
   • this file also does not exist yet and must be wired into the roadmap stack
7. update the file-review and proof ledgers so both new artifacts have explicit ownership instead of being implied

Deleted File Accounting

Family accounting rule:

• deleted file count must still reconcile to 8
• split rows above are claim clusters, not extra deleted files
• the closeout note must still name all eight deleted files and their final disposition

Execution Phases

Phase 0: Freeze The Family Matrix

Goal:

• write the per-file disposition into the execution artifact before code work

Exit:

• all eight deleted rows have an initial disposition and proof owner

Phase 1: Confirm Mirrored Rows

Goal:

• verify the mirrored now rows against current runtime proof so nobody hand-waves them later

Actions:

• cite the exact runtime test owners for:
  • useSlateSelector
  • useSlateWithV
  • current hook surface
  • mounted window/root helper surface
  • callback partition

Exit:

• mirrored rows are named precisely in the family closeout note

Phase 2: Restore The Uncovered Public-Value Rows

Goal:

• add one focused contract file for the still-open React-family value

Target file:

• /Users/zbeyens/git/slate-v2/packages/slate-react/test/surface-contract.tsx

Execution rule:

• surface-contract.tsx starts self-contained
• only extract shared helpers from runtime.tsx if the second restored case would otherwise duplicate the same mount/setup logic again

Required restored cases:

1. ReactEditor.focus initializes selection from null and stays safe mid-transform without DOM target drift or toDOMNode failures
2. ReactEditor.focus does not trigger onValueChange
3. the structured render surface keeps element mount identity stable across splitNodes
4. the structured render surface keeps mount identity stable across mergeNodes
5. Editable defaults translate=\"no\" and still allows override
6. useSelected stays stable when the selected element path shifts after structural edits

Constraint:

• do not recreate the deleted chunking internals just to satisfy old tests
• do not smuggle the dead decorate prop back in; use projections only where that value already exists in the live package

Exit:

• /Users/zbeyens/git/slate-v2/packages/slate-react/test/surface-contract.tsx exists
• every restore now row has a direct current proof owner

Phase 3: Explicitly Cut The Dead Residue

Goal:

• close the rows that no longer belong instead of leaving them vague

Explicit skips:

• packages/slate-react/test/chunking.spec.ts
• chunking-only branches inside use-selected.spec.tsx
• exact decorate API parity from decorations.spec.tsx
• packages/slate-react/test/tsconfig.json

Required note:

• explain that the current architecture owns selection/path stability and rendering behavior through the structured surface and projection store, but does not own the deleted chunk-tree reconcile module or the old decorate prop

Phase 4: Ledger And Closeout Sync

Goal:

• make the closure auditable from the roadmap stack

Required artifacts:

• docs/slate-v2/master-roadmap.md
• docs/slate-v2/overview.md
• docs/slate-v2/release-file-review-ledger.md
• docs/slate-v2/true-slate-rc-proof-ledger.md
• docs/plans/2026-04-09-slate-v2-slate-react-deleted-test-family-closure.md

Required ownership updates:

• surface-contract.tsx enters true-slate-rc-proof-ledger.md as class public surface
• the family closeout note enters the roadmap stack as class maintainer context
• runtime.tsx remains the owner only for the rows it already mirrors

Closeout note must say:

• what was already mirrored
• what was restored now
• what was explicitly skipped
• why no more packages/slate-react/test/** restore work is needed

Batch-exit rule:

• after the family closes, refresh:
  • master-roadmap.md
  • overview.md
  • release-file-review-ledger.md
  • true-slate-rc-proof-ledger.md

Acceptance Criteria

1. all eight deleted packages/slate-react/test/** rows are classified in one place with no vague residue
2. the live public-value rows are owned by either:
   • packages/slate-react/test/runtime.tsx
   • packages/slate-react/test/surface-contract.tsx
3. both new artifacts are created and wired into the ledger/proof stack:
   • packages/slate-react/test/surface-contract.tsx
   • docs/plans/2026-04-09-slate-v2-slate-react-deleted-test-family-closure.md
4. chunking.spec.ts is not resurrected and is instead cut explicitly with rationale
5. tsconfig.json is closed explicitly instead of being left as ghost harness debt
6. the ledger stack changes from “remaining open deletion review includes slate-react” to “slate-react deleted-family bucket closed”
7. the roadmap stack is refreshed per the batch-exit rule
8. the execution batch ends with fresh same-turn evidence, not doc theater

Verification

Direct family proof must include:

• yarn workspace slate-react run test
• yarn lint:typescript

Cross-package regression evidence:

• yarn test:custom

Escalate verification if source outside slate-react changes:

• yarn test:mocha
• yarn workspace slate-dom test

Planning artifact verification:

• bunx prettier --check docs/plans/2026-04-09-slate-react-deleted-test-family-closure-consensus-plan.md

ADR

Decision

Close the Slate React deleted test-family bucket with one focused current contract file plus the existing broad runtime proof.

Drivers

• exact deleted-row accounting matters
• hook/editable/structured-rendering public behavior still matters
• chunk-tree internals do not

Alternatives Considered

• extract shared slate-react test helpers first, then split restored proof
• widen runtime.tsx only
• resurrect the deleted file tree

Why Chosen

• best mix of readable proof, honest cuts, and maintainers-facing accounting

Consequences

• one new focused React-family proof file appears
• runtime.tsx stays the owner for already-restored broad runtime coverage
• dead chunking internals are cut plainly instead of haunting the ledger

Follow-Ups

• execute this batch before packages/slate-history/test/**
• reconsolidate the roadmap stack after the family closes

Agent Roster And Staffing

Available useful agent types for execution:

• planner
• architect
• critic
• executor
• researcher
• verifier

Ralph Path

Use $ralph when you want one controlled batch:

1. freeze the matrix
2. restore surface-contract.tsx
3. touch package code only if the restored proof fails honestly
4. sync ledgers and closeout note
5. refresh master-roadmap.md and overview.md
6. rerun verification

Suggested reasoning:

• high for the first restore pass
• medium for ledger/doc sync

Team Path

Use $team when you want parallel excavation:

• lane 1: mirrored-row evidence extraction from runtime.tsx
• lane 2: surface-contract.tsx restore implementation
• lane 3: ledger + closeout artifact drafting
• main thread: integration, roadmap refresh, and final verification

Suggested launch hint:

omx team 3:executor "Execute the Slate React deleted test-family closure in /Users/zbeyens/git/slate-v2 using the approved plan at /Users/zbeyens/git/plate-2/docs/plans/2026-04-09-slate-react-deleted-test-family-closure-consensus-plan.md. Freeze the eight-row family matrix first, restore the uncovered public-value rows into packages/slate-react/test/surface-contract.tsx, keep runtime.tsx as proof owner for already mirrored rows, explicitly skip dead chunking and test-local tsconfig residue, sync the ledgers plus /Users/zbeyens/git/plate-2/docs/slate-v2/master-roadmap.md and /Users/zbeyens/git/plate-2/docs/slate-v2/overview.md, and stop only with fresh evidence from yarn workspace slate-react run test, yarn test:custom, and yarn lint:typescript."

Team Verification Path

1. merge the mirrored-row evidence notes before code review
2. verify the new contract file stays limited to restored public-value claims
3. rerun the required commands from the integration thread
4. refuse closure if any deleted row still lacks a final disposition