Issues:
- We have limited QA human bandwidth.
- Pull requests are bottlenecking in a "needs QA review" state.
- QA is bogged down with manual/regression testing.
- Communication and the hand-offs between the engineers, seit, and qa can be greatly improved.
Here is the outline of QA experiments/action-items we are going to try:
- Default: QA / SEIT team reviews change
- JIRA: Write non-techie actionable "Acceptance Criteria" in card comments
- Steps-to-repeat are strongly suggested
- A line to give a little context might be good
- Sanity check / smoke test
- Load the server
- Tail the logs
- Look for obvious breaks: icons missing, errors in JS console, errors in log, etc
- JIRA: Move ticket to "Test Queue"
- Notify individual QA/SEIT that review will be necessary
- Walk QA/SEIT through specific change so testing may be localized
- Be persistant, but respectful to get QAs time.
- Make it easy for her/him to say "+1"
- QA sign-off should be done in hours not days
- QA and authors should explain why story is in QA during morning standup.
- Default: Dev team reviews change
- Assumes adaquete testing coverage in place
- Possible layers of testing: unit, functional, integration, feature
- Adequate as judged by team, must be more than "none"
- PR: If a developer suggests QA assistance, author honors that request
- This assumes critical analysis of code has been exhausted
- Author uses process outlined in "UI / UX Change"
@pricees 👍