On 19 Jun 2006, Adelle Hartley <[EMAIL PROTECTED]> wrote: > Surely, for any interpreted language, it ought to be possible to build a > version of the interpreter that records which parts of the byte code were or > weren't executed?
It's possible to measure which instructions were executed or not; it's harder to determine whether they were *exercised*. Suppose we have a test suite that runs the program through all representative states, but makes no assertions whatsoever about the output. We might have 100% coverage but won't actually catch any bugs other than those that cause the program to abort. Mutation testing as I understand it a way to assess "how good is our test suite?" - ie "what is the % chance that it will catch any lurking bugs". One way to do this is to intentionally put in a bug and see if the test suite catches it; iterate this many times and see how many you caught. > Mutation seems kind of hit or miss. The polite word is "probabilistic" :-) -- Martin _______________________________________________ coders mailing list coders@slug.org.au http://lists.slug.org.au/listinfo/coders
