Multi-rooftop phase 2: inventory ownership and tracked transfers (Gitea #397) #668

New Issue

2026-09-10T07:38:15Z

gambit-admin commented

2026-09-10 07:38:15 +00:00

Parent: Gitea #397 — Cards on the Harbor, Rockwall headquarters + Burleson branch.
Depends on: the headquarters/store-directory phase on codex/397-rooftop-locations.

Deliver a complete inventory-ownership and exact-copy transfer workflow before customer activation:

Add durable store ownership separately from shelf/bin location_code; tenant/store foreign keys prevent cross-account links.
Explicit audited initial stock enrollment. Unassigned stock stays visible; do not guess that all historic inventory belongs to HQ.
Scope all intake/import/buy/trade upserts and inventory grouping by store, so Rockwall A1 and Burleson A1 never merge.
Company/store availability including available, reserved, in-transit and unassigned buckets.
Owner-controlled request/reserve -> dispatch -> receive transfers, linked to destination customer demand when supplied. Before-dispatch cancel releases holds; after-dispatch exceptions require a recorded return or resolution.
Preserve GCN, acquisition date, effective original acquisition cost and provenance. Snapshot source aggregate cost when a copy uses fallback cost. No transfer revenue or changes to lifetime units sold.
Every stock writer must honor transfer reservations: canonical transactions, /sell/cart, legacy trade finalization, shows, inventory edits, imports, public reservations and POS webhooks. Explicitly reject unsupported legacy paths for enrolled accounts.
Stable lock ordering, expected versions and idempotency keys. Same exact unit cannot be transferred or sold concurrently.

Acceptance evidence: real PostgreSQL sale/transfer races; timeout/retry of dispatch and receipt moves stock once; quantity and acquisition-value conservation; source/destination and vendor permission checks; protected/listed/show-reserved stock rejected; before-dispatch cancellation and after-dispatch exceptions tested; usable owner transfer screen.

Initial pilot proposal: exact GCN inventory, all-or-nothing receipt, sale after receipt. Bulk/custom units, partial shipments and direct fulfillment before receipt are deferred unless Rudy makes them launch requirements. Attribution and staff-scope decisions remain for Rudy; the parent rollout plan records the recommended defaults.

Do not mark Gitea #397 or the customer pilot complete when only this phase is green. Checkout attribution, staff access and reporting remain a separate gate.

Gambit standard for every storefront business (Rudy, September 10, 2026)

This is the common Gambit account/location model for all storefronts we sign, from one store to 100 or more. Cards on the Harbor is the first customer rollout, not a special-case product. One Vendor owns all StoreLocation records, with one active HQ; no separate account per branch, numbered store columns, customer-name logic or fixed location-count ceiling.

The same operational paths must pass acceptance at one, two and 100 locations and remain isolated across business accounts. A directory-only 101-store check is not proof of 100-store inventory, permissions, checkout or reporting readiness. Record representative stock/staff/transaction volume, concurrent workload, response-time targets and measured results before making that capacity claim.

Source contract: docs/design/multi-rooftop-rollout.md on codex/397-rooftop-locations.

Scale acceptance: company/store availability must query within tenant and authorized stores with bounded pages and server-side summaries. Transfers reference source/destination directly; never configure every pair of stores. Test concurrent independent transfers and competing sale/transfer requests at 100 stores, including destinations beyond page one, shared shelf codes, stock/cost conservation and no permission leaks. A single-store business uses the same stock ownership model; expansion adds a branch without changing account type or moving existing stock.

Parent: Gitea #397 — Cards on the Harbor, Rockwall headquarters + Burleson branch. Depends on: the headquarters/store-directory phase on codex/397-rooftop-locations. Deliver a complete inventory-ownership and exact-copy transfer workflow before customer activation: - Add durable store ownership separately from shelf/bin location_code; tenant/store foreign keys prevent cross-account links. - Explicit audited initial stock enrollment. Unassigned stock stays visible; do not guess that all historic inventory belongs to HQ. - Scope all intake/import/buy/trade upserts and inventory grouping by store, so Rockwall A1 and Burleson A1 never merge. - Company/store availability including available, reserved, in-transit and unassigned buckets. - Owner-controlled request/reserve -> dispatch -> receive transfers, linked to destination customer demand when supplied. Before-dispatch cancel releases holds; after-dispatch exceptions require a recorded return or resolution. - Preserve GCN, acquisition date, effective original acquisition cost and provenance. Snapshot source aggregate cost when a copy uses fallback cost. No transfer revenue or changes to lifetime units sold. - Every stock writer must honor transfer reservations: canonical transactions, /sell/cart, legacy trade finalization, shows, inventory edits, imports, public reservations and POS webhooks. Explicitly reject unsupported legacy paths for enrolled accounts. - Stable lock ordering, expected versions and idempotency keys. Same exact unit cannot be transferred or sold concurrently. Acceptance evidence: real PostgreSQL sale/transfer races; timeout/retry of dispatch and receipt moves stock once; quantity and acquisition-value conservation; source/destination and vendor permission checks; protected/listed/show-reserved stock rejected; before-dispatch cancellation and after-dispatch exceptions tested; usable owner transfer screen. Initial pilot proposal: exact GCN inventory, all-or-nothing receipt, sale after receipt. Bulk/custom units, partial shipments and direct fulfillment before receipt are deferred unless Rudy makes them launch requirements. Attribution and staff-scope decisions remain for Rudy; the parent rollout plan records the recommended defaults. Do not mark Gitea #397 or the customer pilot complete when only this phase is green. Checkout attribution, staff access and reporting remain a separate gate. ## Gambit standard for every storefront business (Rudy, September 10, 2026) This is the common Gambit account/location model for all storefronts we sign, from one store to 100 or more. Cards on the Harbor is the first customer rollout, not a special-case product. One Vendor owns all StoreLocation records, with one active HQ; no separate account per branch, numbered store columns, customer-name logic or fixed location-count ceiling. The same operational paths must pass acceptance at one, two and 100 locations and remain isolated across business accounts. A directory-only 101-store check is not proof of 100-store inventory, permissions, checkout or reporting readiness. Record representative stock/staff/transaction volume, concurrent workload, response-time targets and measured results before making that capacity claim. Source contract: docs/design/multi-rooftop-rollout.md on codex/397-rooftop-locations. Scale acceptance: company/store availability must query within tenant and authorized stores with bounded pages and server-side summaries. Transfers reference source/destination directly; never configure every pair of stores. Test concurrent independent transfers and competing sale/transfer requests at 100 stores, including destinations beyond page one, shared shelf codes, stock/cost conservation and no permission leaks. A single-store business uses the same stock ownership model; expansion adds a branch without changing account type or moving existing stock.

gambit-admin added the feature ui api labels 2026-09-10 07:38:15 +00:00

gambit-admin referenced this issue

2026-09-10 07:38:49 +00:00

Multi-rooftop phase 3: standard onboarding, store sales, access and rollout acceptance (Gitea #397) #669

gambit-admin referenced this issue

2026-09-10 08:36:55 +00:00

Gambit standard: one business account with one to 100+ store locations #397

gambit-admin added the claimed:rudy2 label 2026-10-02 15:58:39 +00:00

Sign in to join this conversation.