chin fan soProgram Manager| Galaxy EntertainmentTaipa, Macao
I found a very difficult (if not the most) part of Agile is to persuade the QA team to re-test the deliverables in the previous sprints, since Agile encourage feature after feature to be built on the same functional area, so it will have to be test and test again for every sprint. If it is not a function that can easily automated, it is very hard to convince the QA team to re-test it while they don't need to do it in waterfall approach (which the whole function delviered at once).
How do you deal with this problem?
Any comment will be greatly appreciate. Saving Changes...
It's usually better to have cross-functional teams responsible for both development and QA. That way the priorities for dev and QA are aligned, collaboration is straightforward and each iteration delivers a potentially releasable, tested increment.
More automation is ultimately what you need. Your regression testing should by definition be possible to automate since regression testing essentially answers the question "did it change?" and not "is it right?". Having a curated set of inputs, tests and expected results ought to solve that problem. The QA efforts should be focused on test curation and defect analysis rather than test execution. Saving Changes...
Sergio Luis ConteHelping to create solutions for everyone| Worldwide based OrganizationsBuenos Aires, Argentina
Test must be done in the same sprint because nothing is done after is done. If you are retesting something is not working in your process. That´s true except you are talking about maintenance activities where you have to retest previous components due to you make a fix (any type) on those components. Saving Changes...
Unfortunately, without automation for regression testing, you are unlikely to get sufficient coverage even if you have a willing QA team to be able to feel confident that newly added features haven't broken previously delivered functionality.
In such situations, it is not uncommon to adopt a "water-scrum-fall" or "scrum-fall" pattern, where comprehensive end-to-end testing gets pushed to the end of a release, resulting in a higher cost of rework when regression defects are identified many sprints after they were injected.
A shift to test-first/driven development and a real commitment to continuous automated testing is needed to overcome this.
