9.3 KiB
Feature test-case catalog
Purpose and status
This folder contains feature-by-feature acceptance and regression checklists for Freshyo's customer mobile/web apps, regional admin mobile/web apps, superadmin web app, backend, cache/jobs, and legacy fallback surfaces.
All cases are proposed and NOT RUN. A listed expected result is a requirement to verify, not a claim that the implementation currently satisfies it. Known review defects are intentionally covered. Documentation creation does not authorize running tests, sending notifications, charging payments, deleting data, or applying migrations.
Inventory: 45 feature checklists and 804 uniquely identified cases, plus this index. Features are grouped by business responsibility; closely related operations share a checklist (for example stock, offer flags and suspension). Reuse the shared boundary/security matrix rather than assuming a finite list can cover every possible input.
Sources: current routers/screens/helpers, existing API tests, sanity plan, and branch review. Older sanity-plan statements that coupons ignore regions/SKUs are superseded here by the newer coupon requirements. Vendor snippets and reviews remain conditional/legacy coverage; online gateway cases are readiness coverage until implementation is enabled.
How to use each checklist
- Each case has a stable ID, priority, concrete action/input, and observable expected result.
- P0: release-critical authorization, financial integrity, or core purchasing. P1: normal functional/edge regression. P2: resilience, presentation, compatibility, or conditional feature coverage. These are test priorities, not defect severities.
- Run happy paths through the UI; validate persistence and security through authoritative API/read-only database evidence. Component tests alone do not prove end-to-end behavior.
- Execute API checks even if a control is hidden in the UI. Repeat on every relevant client; web/native are separate implementations.
- Record
PASS,FAIL,BLOCKED,NOT RUN, orN/Ain a separate run report.N/Aneeds a reason;BLOCKEDis not a pass. Do not check off cases solely because a toast appeared. - For every rejected mutation, verify no partial row changes, sibling changes, image deletion, or unrelated-region writes.
- For every money case, compare cart, checkout, success, persisted order, customer detail, and admin detail:
goods − stored discount + delivery = payableper sub-order and in aggregate. Delivery is charged once per checkout group; no negative/nonfinite monetary values. - For cache cases, record authoritative value, region/bound, served version, and elapsed time. Use a bounded publication deadline (initial staging baseline: 120 seconds), not unbounded sleeps.
- Treat policy-dependent cases as
BLOCKED: POLICYuntil agreed. Never infer a product decision from a current bug.
Shared fixtures
Use isolated staging with dedicated DB/KV/assets/queues and sandbox recipients. Verify the actual API host before any mutation; do not use production credentials. Read secrets from approved environment configuration, never this documentation.
| Fixture | Required setup |
|---|---|
| Regions | A and B; A1/A2 bounds with priorities 1/2 and different fees; B1; valid interior, exterior, edge and overlap coordinates. |
| Identities | Anonymous, customers U1/U2, new customer U0, suspended disposable customer, regional admins A/B, supported staff roles, superadmin. |
| Products | A-only P with two SKUs, B-only Q, global G with independent A/B prices; regular-only R; flash F with a distinct price; combo C and combo-only component K. |
| Availability | Available, OOS, globally suspended, regionally suspended, soft-deleted variants; preserve a historical order referencing one. |
| Slots | Two valid A slots S1/S2, B slot, inactive/frozen/full/past slots, and midnight-boundary times. |
| Addresses | U1 in A1/A2/B1; U2-only address; boundless legacy address; default and sole-address fixtures; coordinate-free address. |
| Coupons | Regional A/B and global; percentage/flat; targeted/whole-cart; capped, min-spend, exclusive, first-order, limited-use, user-only, expired/invalidated, reserved. |
| Orders | Pending COD, packaged, verified, delivered, cancelled, multi-slot group; payment/refund fixtures only in approved sandbox. |
| Money | T = free threshold, D = fee; test T−0.01, T, T+0.01. Example goods: ₹100 targeted + ₹900 unrelated. Separate regular/flash values. |
| Content | Tags, groups, active/inactive banners, images, supported notification links, complaints and incident fixtures. |
Use unique run prefixes. Do not delete regions for cleanup: region deletion is unsupported. Never run migrations as test setup. Keep shared fixtures intact; clean only approved disposable data. Use time-relative slots/expiry dates with margins, except deliberate exact-boundary cases.
Feature index
| Area | Checklist |
|---|---|
| Identity | Customer authentication, Profile/account lifecycle, Staff authentication, Staff management |
| Geography | Regions, Bounds, Location and switching, Constants |
| Catalog administration | Regional stores, Global stores, Regional products, Global products/elevation, Prices/stock/suspension, Combos |
| Merchandising | Catalog/search, Tags/order, Groups, Banners |
| Purchasing | Slots, Cart, Addresses, Checkout, Pricing, Flash |
| Coupons | Administration, Eligibility/stacking, Reserved/cancellation coupons |
| Order operations | Placement/history, Fulfillment, Cancellation, Edits/rebalance, Delivery sequence/maps, Payment/refund readiness |
| Support | Complaints, Customer operations/incidents, Notifications |
| Infrastructure | Uploads, Cache publication, Jobs/health, Security, Recovery/platform/accessibility |
| Conditional/legacy | Vendor snippets, Reviews, Fallback/legacy pages |
| Approval-required provider integration | OTP authentication (isolated provider sandbox only; excluded from default password regression) |
Policy checkpoints
- Currency rounding and residual allocation: choose a consistent minor-unit policy; don't assert an invented rounding mode.
- Percentage stacking: confirm whether successive percentages use original or remaining eligible value. Regardless, targets, caps, nonnegative totals, and client/server agreement are mandatory.
- Global coupon targeting regional products: confirm intended authoring restrictions; never permit a regional author to attach another region's private SKU.
- Combo component availability, group-to-slot snapshot/live membership, global-name uniqueness/case normalization, and account-switch cart behavior need explicit decisions where not specified.
- No general minimum order value is required by the current policy. Coupon
minOrderis independent from free-delivery thresholds. - UI duplicate-submit prevention is required. Server order-idempotency was previously deferred: record concurrent API behavior and ambiguous-retry exposure, not an assumed guaranteed deduplication.
- Define concurrency guarantees for capacity, default-address updates, limited coupons, and reserved redemption. Financial/one-time invariants must not silently be weakened to match observed races.
Known-defect regression mapping
| Review issue | Cases |
|---|---|
| Web discovery input mismatch | CFG-01, CAT-01, CACHE-01 |
| Global SKU ID stripped on edit | GP-03, GP-04 |
| Overlapping coupons/negative sub-orders | CPA-14, MONEY-12 |
| Uncapped mobile preview | CPA-15, MONEY-13 |
| Regional suspension lost on reopen | STOCK-07, STOCK-08 |
| Global SKU rejected by bulk prices | STOCK-03, STOCK-04 |
| Targeted minOrder uses whole cart in UI | CPA-07, CPA-08 |
Run-report template
Run ID / date / code revision:
Environment / verified API host / client build:
Case ID / client / identity / region / bound:
Fixture IDs and relevant pre-state:
Actions and exact inputs:
Expected / actual:
Status and reason:
API status/body, persisted values, versions, timestamps:
Screenshot/log links (redacted):
Cleanup / retained evidence / linked defect:
Recommended first pass: AUTH-01, CFG-01, LOC-06, GP-03, STOCK-03, STOCK-07, SLOT-01, CART-01, ADDR-01, CHECK-01, MONEY-01, CPA-01, CPA-07, CPA-14, ORDER-01, FUL-01, CANCEL-01, SEC-01, CACHE-01. Then execute all P0 cases before release and the remaining applicable cases in the full regression run.