20191121 - BREWT - Mutation Testing: Opportunities and Pitfalls, Ali Parsai, @AliParsai

by Thierry de Pauw on

#brewt

Mutation Testing: Opportunities and Pitfalls, Ali Parsai, @AliParsai

Do I really have to test? No. But your code quality will suffer. @AliParsai

ESA Mars Lander blew because of an integer overflow => 250 millions burned

Better Test Quality => Fewer Bugs

Tests run against your software to be sure the software is right.
But we want to know the quality of the tests
=> inject a bug into the software
=> your software turns into a mutant
You want your tests to detect that bug
If the test detected the bug => killed the mutant
If it did not detect the bug => mutant survived

Java: very developed
C/C++: is almost there

If you have a coverage of 30% don't do mutation testing, only start when you have a coverage of 90%.

When running mutation testing on a 30% coverage you will miss lots of mutants that are not covered by the tests.