Why are log probabilities often used when implementing Naive Bayes?

A. They guarantee perfect calibration B. They remove the conditional-independence assumption C. They turn products into sums and reduce numerical underflow D. They change all priors to equal values