Error guessing
Just as you need to choose the values of the inputs you will use for black-box testing, you also need to select the values for white-box testing. These also require equivalence partitioning and the careful choice of examples to represent each partition. As ever, the requirement here is to challenge the code as much as possible and try to provoke errors.
The lists from Chapter 5, Black-Box Functional Testing, are helpful. Instead of integers being entered into a web page or sent as part of an API, in this case, they will be passed between functions, but the same range of interesting cases still applies. How does the function handle negative numbers, decimals, or large values? Likewise, strings and other data types.
Watch out for historical anomalies in the code. For instance, sometimes data or configuration is stored one way up to a certain release and then a different way after that. Sometimes it is worthwhile to go back and migrate the old data to the new system...
