SFScon18 - Kurt Ranalter - Leonhard Holzer - API level regression testing: a field report

SFScon18 - Kurt Ranalter - Leonhard Holzer - API level regression testing: a field report

• 1.
  API Level Regression Testing:a Field Report Leonhard Holzer & Kurt Ranalter SFScon 2018
• 2.
  Mission statement • Softwarecompany from Bruneck-Brunico • BI, quality control & production planning tools • Customers: automotive industry, local business • Agile, CI/CD: continuous integration & delivery • One of our aims is to deliver high quality SW • Testing is crucial to avoid breaking functionality • But testing is not for free, it costs time and effort • Apit: our API level regression testing framework API level regression testing 2SFScon 2018
• 3.
  Example application API levelregression testing 3SFScon 2018
• 4.
  Issues & challenges •Page dependencies • PNL & balance → cash flow • Accruals → PNL & balance • Input dependencies • Drilldowns & data aggregation • Aggregated row ↔ data row (→) depending on reference column • TY column ↔ month column (→) based on month, working days API level regression testing 4SFScon 2018
• 5.
  Behind the scenes Postrequest Get request SFScon 2018 API level regression testing 5
• 6.
  Architecture & testing •Focus on testing business logic • End-to-end testing not really an option • Usual solution: unit & integration tests • Requires extensive setup and mocking • Why not make use of REST API? • Simulate usage as in end-to-end testing • Business logic considered as black box • Not a substitute but complementary SFScon 2018 API level regression testing 6
• 7.
  Apit DSL forrequests Post request Apit test case SFScon 2018 API level regression testing 7
• 8.
  How does itwork? • In a nutshell 1. Add framework as library 2. Read and parse config files 3. Run as JUnit5 dynamic tests • Build pipeline 1. DB/auth as docker container 2. Tests just run as maven goals API level regression testing 8SFScon 2018
• 9.
  Concluding remarks • Designdecisions & supported features • Exploit framework to generate necessary test data • Actual/expected as JSON (WIP: git to track changes) • Payloads from file instead of coding it into config • Parameters for testing in multiple environments • Quick & easy to use but still powerful • No need for mocking, relatively simple to maintain • Good integration with respect to our build pipeline API level regression testing 9SFScon 2018
• 10.
  That‘s it, thanks! www.abuscom.com [email protected] APIlevel regression testing 10SFScon 2018