- Software engineering research needs a database of reliably reproducable real faults for many reasons.
What research doesn’t have is time for is actually collecting this database. So I spent
far too much time dealing with idiosyncrasies of build systems, library version compatibility,
and the dreaded
CLASSPATHsetting, with my collegue René Just. Unfortunately for you, we did it in Perl. It still beats dealing with
gitcommit SHAs though, so enjoy using it.
- Dependent test detector
- Have you ever run your tests in a different order and got different results? No? Well, that’s because you aren’t a practicioner – you should go ask one. I spent some time on Sai Zhang’s tool and added a program analysis that eliminated the need to check every order to determine whether running a certain reordering of tests may change the results from the original order.