Just a (probably naive) question to improve my understanding of how to set up the fuzzers: I'm surprised we run an additional test if res == 0, as opposed to branching to a separate test on the value of res.

How does that work / why do we want to run both?

How would this compare to code like:

if (args->res == 0) {
    H3_EXPORT(constructCell)
    (args->res, args->baseCellNumber, NULL, &out);
} else {
    H3_EXPORT(constructCell)
    (args->res, args->baseCellNumber, args->digits, &out);
}

Or, is the idea that the res == 0 branch provides a more targeted test that's more likely to find issues quickly because it is reducing the search space?

I think if res is 0 and a non-NULL pointer is passed for digits it should still be accepted. The inverse is not true however, since the caller must provide an array of length res.

I'm updating this to be clearer about why we are testing different digits lengths.

Maybe this is just auto-formatting, but I find this hard to read compared with opening the parens on the previous line

This is unfortunately how the auto-formatter wants to do it