[I] Backend e2e specs assume a freshly seeded instance and fail when one is reused [fineract-backoffice-ui]

["Aman-Mittal (via GitHub)" <gi...@apache.org>]

Aman-Mittal opened a new issue, #664:
URL: https://github.com/apache/fineract-backoffice-ui/issues/664

   Running the full `backend` project end to end against a Fineract container that had already served several runs produced two failures out of 117:
   
   ```
   117 tests: 115 passed, 2 failed (13.8m)
   
   1) rbac-office-scoped-user.spec.ts:161 — is not offered an edit it would be refused for
   2) report-parameter-backend.spec.ts:67  — changing Office changes the Client Listing row set
   ```
   
   Neither is an application defect. **Both pass on a fresh stack, which is why CI is green** — CI starts a clean container per run. They fail for anyone running the backend project twice against the same one, which is the normal local workflow and the documented one in `DOCS/E2E_TESTING.md`.
   
   Both have the same cause: the spec seeds a record, then looks for it on whichever page the table renders first. Seeded records accumulate.
   
   ## Offices
   
   `.data-table` here is `localLogic` with the shared ten-row default. The instance had 32 offices, so a branch created moments earlier was not on the first page:
   
   ```
   Locator: locator('.data-table').getByText('E2EScoped Branch qw6uro').first()
   Expected: visible
   Error: element(s) not found
   ```
   
   ## Client Listing report
   
   This one cannot be fixed by paging at all. The spec already bumps the paginator to 100, but `pageSizeOptions` is `[5, 10, 25, 100]` — 100 is the ceiling — and Head Office had passed 118 clients, so the seeded client sat beyond any page a test could turn to.
   
   ## The general shape
   
   Any assertion of the form "the record I just seeded is visible in this table" is a time bomb on a long-lived instance, because the haystack only grows. Two durable fixes, depending on the screen:
   
   - **Filter by the seeded name**, which is unique per run, so pagination stops mattering. Preferred over widening the page size — a bigger page size is the same bug deferred.
   - **Assert on the response rather than the rendered page**, where what the spec is really about is which rows came back rather than which rows were painted.
   
   Fixed for these two in #661. Other specs that seed into a shared list and then assert on the rendered table may have the same latent problem; this issue is the place to record further instances as they surface.
   
   Related: #653 records a separate class of backend-coverage gap.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: issues-unsubscribe@fineract.apache.org.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org