"Test first bug fixing" is probably the highest ROI activity in the entire testing lifecycle. There is no better way to understand a bug than being forced to reproduce it in a minimal, isolated environment.
The challenge I see most often in industry is that reproduction in a test environment is often 90% of the work. If your system requires a massive integration test just to reproduce a simple logic bug, it’s a sign that your testing pyramid has inverted into a "testing ice cream cone."
pfdietz•13h ago
What can help there is automatic test case minimization. This requires being able to replay test inputs while minimizing them, so there must be no nondeterminism in the system under test.
alphadatavault•13h ago
The challenge I see most often in industry is that reproduction in a test environment is often 90% of the work. If your system requires a massive integration test just to reproduce a simple logic bug, it’s a sign that your testing pyramid has inverted into a "testing ice cream cone."
pfdietz•13h ago