Even after 6 years in the field, I ask myself this question all the time. Once I tested something that worked like a shared document with a version history. A specific action would create a new version of it. The original must never change.
I created different documents and tried to change them in different ways, and it was all working pretty well. I was checking on all the obvious places, but the bugs are usually not there.
I know that it isn’t realistic to verify that there are no bugs in the software. There are always the obvious ones, but it takes real critical thinking and understanding of the software to find the ones lurking in the shadows.
I like to hold myself to high standards, and over the years I’ve learned where to look for them. One of my main influences is the book “Explore It!” by Elisabeth Hendrickson.
As an exploratory tester I need to find every piece of data I can create, read, update or delete, and ask several questions about it:
Where does it live? The UI, the database, API responses, messages sent to other services, the logs?
What values can it take? What values shouldn’t it take?
What can trigger a change in the state?
Back to the document. I was missing one variable that was not mentioned anywhere – what if I created version 3 out of version 2, would I be able to change the original one then?
Yes, the original one could be changed. This is where the final question paid off. Catching that before it was released prevented the possibility of historical data being altered.
This is just a small example of why defining test coverage can be tricky. I’m curious to hear if you have stories with hard-to-find bugs.
Top comments (0)