Assert the merged navigation tree after the package split equals the pre-split tree (objectstack#14553 ruling C half) #1456

Activity

1. os-project-manager

   added

   pm:queueReady for the PM dispatch loopReady for the PM dispatch loop

   on Sep 2, 2026
2. hotlong commented on Sep 2, 2026

   

   hotlong

   on Sep 2, 2026 – with Claude

   Contributor

   More actions

   First-touch grading → pm:blocked + type Task. ⛔ Not pm:queue: the thing this card compares against does not exist yet.

   Graded by the repo:hotcrm seat, R18, as the card asks. ⛔ The ruling is not narrowed or re-litigated — this is sequencing, not scope.

   The premise check

   The card says to compose the split packages and deep-compare the merged navigation tree against the pre-split tree, describing "hotcrm's navigation conversion (hotcrm#1449, 17 nodes)". Measured on origin/main @ c73b496b:

   top level:  .changeset .claude .github AGENTS.md apps assets content docs e2e
               objectstack.config.ts objectstack.manifest.json package.json scripts
               src test tsconfig.json vitest.config.ts
   packages/:  (does not exist)
   
   composeStacks / navigationContributions — every hit:
     docs/architecture/module-split-plan.md
     docs/architecture/module-split-inventory.json
     .changeset/module-boundary-inventory-and-split-plan.md
   

   ⇒ hotcrm is still one package. Those two identifiers appear only in plan documents, never in code. And #1449 is not a conversion — its PR is docs(architecture): module-boundary inventory and split plan for the ADR-0130 multi-package artifact, and it landed exactly three files: a plan, an inventory JSON, and a changeset. The 17 nodes are planned contributions, ⛔ not converted ones.

   ⇒ There is nothing to compose and no post-split tree to compare, so a dispatch today would end at premise_still_valid: false before writing a line. Blocking it is the honest state, and it costs nothing: the ruling stands and fires the moment the split lands.

   ⭐ One thing I checked because it would have changed the grading, and it does not

   My first read was that this card is urgent rather than blocked — that the pre-split baseline is a perishable asset that vanishes when the split lands, so someone must capture it now even if the comparison waits.

   That is wrong, and the card had already closed it: it offers the baseline as "a fixture captured from the last single-package build, or computed from the single-package definition on the same commit". The second form is reconstructible from git at any time. ⇒ Nothing perishes, and there is no carve-out worth queuing ahead of the split. Recording the check because "capture it before it's gone" is exactly the argument that would have justified splitting this card, and it does not hold.

   Blocker

   Blocked-by: objectstack-ai/objectstack#14122 — the multi-package epic, written into the body so the unlock sweep can see it.

   ⚠️ Named precisely because it matters for whoever unblocks this: no hotcrm-side card tracks the split conversion itself. #1448 (the inventory) is closed, #1449 (the plan) is merged, and nothing in this repo's backlog carries "perform the split". So this card is blocked on work that has a plan and an epic but no consumer-side tracker. ⇒ When the split is scheduled, a hotcrm card for it should be filed and this one re-pointed at it; until then the epic is the most honest blocker available.

   On unblocking

   The acceptance criteria as written are good and need no change — order-sensitive, group-sensitive, plus a positive control that mutates one group to a non-existent id and confirms red. ⭐ That last one is the part that matters most here, because the defect objectstack#14553 measured is silent relocation: without an ablation, a test that merely passes proves nothing about whether it would have caught the relocation.

   ---

   Generated by Claude Code
3. hotlong

   added

   ci/cdCI plumbing and the verification pipelineCI plumbing and the verification pipeline

   pm:blocked

   and removed

   pm:queueReady for the PM dispatch loopReady for the PM dispatch loop

   on Sep 2, 2026
4. hotlong

   added the

   Task

   issue type on Sep 2, 2026
5. objectstack-fleet commented on Sep 25, 2026

   

   objectstack-fleetbot

   on Sep 25, 2026

   Contributor

   More actions

   repo:hotcrm seat, session_01X8U3asekbiC7yWoEPWR4Dg · stock re-triage group 3 (maintainer-confirmed ten-card group; maintainer reply verbatim: 「同意」) · 2026-09-25T02:19Z

   Closed not_planned — nothing left that a separate card should carry

   • The group-typo half is covered by the platform: the published @objectstack/cli@17.4.0 ships dist/utils/nav-contribution-groups.js, a build-time check that reports a missing navigation group.
   • Ordering and grouping are already pinned by test/docs-quick-tour-navigation.test.ts, which Packaging: sales becomes the type: app package and service a type: module — composeStacks(…, { manifest: 'preserve' }), navigation contributions, two-entry packages[] #1907's acceptance (item 4) must teach to read the contribution channel.
   • A literal pre-split/post-split equality would contradict Packaging: sales becomes the type: app package and service a type: module — composeStacks(…, { manifest: 'preserve' }), navigation contributions, two-entry packages[] #1907's accepted delta (nav_my_cases lands last in My Work).

   Hint for #1907's epic PM: compare against the pre-split tree once, and admit that one delta.

   ---

   Generated by Claude Code
6. objectstack-fleet

   removed

   pm:blocked

   on Sep 25, 2026
7. objectstack-fleet

   closed this as not plannedon Sep 25, 2026