NEC-271: P4 · Causal deletes for canvas and base records #344

Merged
nectenda-agent merged 2 commits from worktree-nec-271-causal-deletes-for-canvas-and-base-recor into main 2026-11-08 12:20:18 +00:00
Collaborator

Task: NEC-271 — https://projectron.nerchure.com/tasks/282

NEC-271 proposed storing the deleter's Yjs state vector in each SAFE-A18 tombstone so an edit by a client that knew of a delete could never revive a canvas or base record. Worked through against the code, the vector cannot replace the snapshot: the snapshot is deliberately the file's state, not the document's; a card a peer deletes mid-drag must still come back; removals leave no clock; in the NEC-322 order the link update really is concurrent with the delete; and the deleter's vector cannot say what a later writer knew. It would add a schema bump and migration while closing no accepted case. The user chose to record the finding instead: a change spec with the analysis, and a paragraph in each of the two sync-limitations entries that hold the open re-point case. No code changes. The schema 1→2 upgrade finding made on the way (tombstone snapshot paths not rewritten) is filed as NEC-380 in Backlog.

Spec: docs/changes/NEC-271-causal-deletes/spec.md

Changelog

NONE

Task: NEC-271 — https://projectron.nerchure.com/tasks/282 NEC-271 proposed storing the deleter's Yjs state vector in each SAFE-A18 tombstone so an edit by a client that knew of a delete could never revive a canvas or base record. Worked through against the code, the vector cannot replace the snapshot: the snapshot is deliberately the file's state, not the document's; a card a peer deletes mid-drag must still come back; removals leave no clock; in the NEC-322 order the link update really is concurrent with the delete; and the deleter's vector cannot say what a later writer knew. It would add a schema bump and migration while closing no accepted case. The user chose to record the finding instead: a change spec with the analysis, and a paragraph in each of the two sync-limitations entries that hold the open re-point case. No code changes. The schema 1→2 upgrade finding made on the way (tombstone snapshot paths not rewritten) is filed as NEC-380 in Backlog. Spec: `docs/changes/NEC-271-causal-deletes/spec.md` ## Changelog NONE
NEC-271 proposed storing the deleter's state vector in each tombstone so an
edit by a client that knew of the delete could never revive a canvas or base
record. Worked through against the code it cannot replace the snapshot: the
snapshot is the file's state, not the document's; removals have no clock; in
the NEC-322 order the link update really is concurrent with the delete; and
the vector cannot say what a later writer knew. The spec records the analysis
and the two sync-limitations entries gain a paragraph each, so the next
reader does not re-derive it. No code changes.

Task: NEC-271

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EV8ztXjPJZviB7eW4THywV
Correct two citations in the NEC-271 analysis after review
All checks were successful
Release note / release-note (pull_request) Successful in 17s
CI / build (pull_request) Successful in 3m49s
CI / e2e (pull_request) Successful in 16m6s
CI / promote (pull_request) Has been skipped
CI / build (push) Successful in 4m5s
CI / e2e (push) Successful in 16m4s
CI / promote (push) Successful in 32s
bc657a0183
The drag paragraph now names both rules at work — SAFE-A20 holds the removal
back, SAFE-A18 keeps the moved card — and finding 6 no longer counts the
scratch documents' tombstones, which never leave the scratch document.

Task: NEC-271

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EV8ztXjPJZviB7eW4THywV
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
Nectenda/nectenda!344
No description provided.