Comment by ken

Hmm, isn't their example in 7.0 wrong? The opposite of "a>b" (in C, anyway, AFAIK) isn't "a<=b", because if a==b==NaN both expressions are false (right?).

Just goes to show how hard testing all cases is! How many cases do you need to get full branch coverage on "a>b && c!=25"? I'm thinking 5, but I'm not very sure of that.