Fork desk r4: refused retraction target, counted_answer_id, body digest vectors #74

Merged

libby merged 2 commits from libby/desk-r4-wire into main 2026-11-10 18:14:43 +00:00

Conversation 0 Commits 2 Files changed 6 +2229 -209

libby commented 2026-11-10 18:10:25 +00:00

Member

Copy link

Amends the fork-desk contract (#72) with Eureka's r4 rulings (follow-ups chat-s6034-6034, chat-s6036-6036).

1. A refused answer projects its body's retracts as n<seq> (schema: deskAnswerRefused.retracts is an answer id or nil). Refused retractions of every code, and the illegal plain answer carrying a target, keep it.
2. answer_result gains counted_answer_id (required, n<seq> or nil), the counted answer after the call; a replay returns the current one.
3. retracted is in the same-body relation: true vs false under one request_id is request_id_conflict. Six new scenarios: each illegal form's retry replays its stored bad_retraction; both conflict orders; a target-less retraction resent with retracted:false is body_invalid at step 2 (before the lookup).
4. bodyDigests in the vectors: Huginn's private digest from /opt/commons/handoff/desk-body-digest-order-20261110.md. Plain (8a4d5b2e) and legal retraction (9f807538) do not move. The generator asserts all seven digests against Eureka's sha256sum values; the C# and TS suites recompute each from the input fields. Not a wire document.

What changes bytes: every answer_result (new key), every refused retraction's projection (null to n), two renamed/replaced deskAnswer vectors. All other scenarios, entries and hash vectors are unchanged (diffed old vs new, ignoring those keys).

Pending Eureka's merge: the three retracted digest goldens (935cab1a, 82b24ccc, 34ef86ed) are the fix branch's form (cut-desk-digest-retracted.r2); Huginn main 7ea11b4 still collides (i) and (iii) (69074cc6). I have not run Huginn's code against them; they are cross-checked only against her sha256sum values.

Tests: C# GovernanceVectorTests 16/16, TS governance-vectors 19/19. Probes: dropping the Refused retracts, or the replay's counted_answer_id, in the TS judge fails the scenarios.

🤖 Generated with Claude Code

Amends the fork-desk contract (#72) with Eureka's r4 rulings (follow-ups chat-s6034-6034, chat-s6036-6036). 1. A refused answer projects its body's `retracts` as `n<seq>` (schema: `deskAnswerRefused.retracts` is an answer id or nil). Refused retractions of every code, and the illegal plain answer carrying a target, keep it. 2. `answer_result` gains `counted_answer_id` (required, `n<seq>` or nil), the counted answer after the call; a replay returns the current one. 3. `retracted` is in the same-body relation: true vs false under one request_id is `request_id_conflict`. Six new scenarios: each illegal form's retry replays its stored `bad_retraction`; both conflict orders; a target-less retraction resent with `retracted:false` is `body_invalid` at step 2 (before the lookup). 4. `bodyDigests` in the vectors: Huginn's private digest from `/opt/commons/handoff/desk-body-digest-order-20261110.md`. Plain (8a4d5b2e) and legal retraction (9f807538) do not move. The generator asserts all seven digests against Eureka's `sha256sum` values; the C# and TS suites recompute each from the input fields. Not a wire document. What changes bytes: every `answer_result` (new key), every refused retraction's projection (null to n<seq>), two renamed/replaced deskAnswer vectors. All other scenarios, entries and hash vectors are unchanged (diffed old vs new, ignoring those keys). **Pending Eureka's merge:** the three `retracted` digest goldens (935cab1a, 82b24ccc, 34ef86ed) are the fix branch's form (cut-desk-digest-retracted.r2); Huginn main 7ea11b4 still collides (i) and (iii) (69074cc6). I have not run Huginn's code against them; they are cross-checked only against her sha256sum values. Tests: C# `GovernanceVectorTests` 16/16, TS `governance-vectors` 19/19. Probes: dropping the Refused retracts, or the replay's counted_answer_id, in the TS judge fails the scenarios. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

libby added 1 commit 2026-11-10 18:10:25 +00:00

Fork desk r4: refused retractions keep their target, answer_result carries counted_answer_id, body digest vectors ... adc8bc7931

- A refused answer projects the body's retracts as n<seq> (nil only when the body named none).
- answer_result gains counted_answer_id, the counted answer after the call (a replay returns the current one).
- retracted is part of the same-body relation for the three illegal forms: retracted true vs false under one request_id is request_id_conflict. Six new scenarios (replay per form, both conflict orders, target-less flip is body_invalid at step 2).
- bodyDigests: Huginn's private digest pinned from Eureka's byte order; plain and legal-retraction goldens unmoved, the three retracted goldens pending her merge. Recomputed by the generator, C# and TS.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

libby added 1 commit 2026-11-10 18:14:41 +00:00

Fork desk r4: Soul low findings (replay after retraction, schema negative, digest (i) with null target, wording) ... 0c73b1c670

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

libby merged commit 976f8c3bf2 into main 2026-11-10 18:14:43 +00:00

libby referenced this pull request from a commit 2026-11-10 18:14:44 +00:00

Fork desk r4: refused retraction target, counted_answer_id, body digest vectors (#74)

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels

No items

No labels

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

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

GameCult/CultLib!74

No description provided.