Skip to content

Dataset Scenarios and Practices

This page shows common dataset workflows and maintenance rules.

One Checkout Test for Many Payments

A test case describes checkout flow once. The dataset stores cards, currencies, and expected provider statuses. Manual and automated runs can use the same row keys.

Regional Difference Control

Tax calculation by market stores country, VAT rate, rounding mode, and expected total. When tax logic changes for one market, the team updates one row and sees which cases use it.

Safe Data Changes Before a Release

Before release, a QA lead can pin a case to a dataset revision through the API so late dataset edits do not change an agreed release run. After release, the binding can return to follow_latest.

Manual and Automation Alignment

Automation imports CSV/JSON from pytest parameters, and manual testers link the same dataset to a manual case. Both workflows use the same scenario keys.

Practices

  • Keep row_key stable and readable, such as visa_eur_success.
  • Do not use a dataset as a requirements document; keep it to parameters and expected values.
  • Limit columns to fields used by the case.
  • Fill source_ref when data comes from automation, CSV, or another system.
  • Use pin_revision for release-critical reproducibility.
  • Use follow_latest for active regression data that should pick up new rows.