In MIES I fixed a treewide issue, https://github.com/AllenInstitute/MIES/pull/2153, where I wanted to have CDF empty after the tests. The reasoning was that this avoids bugs where routines rely erroneously on data in CDF.
I started from ~250 failing tests and got down to zero. But I had to run the full test suite too often which took to long.
It would have been useful to have a flag which allows to rerun only the failed tests. ctest also has that 1 and that is quite useful.
In MIES I fixed a treewide issue, https://github.com/AllenInstitute/MIES/pull/2153, where I wanted to have CDF empty after the tests. The reasoning was that this avoids bugs where routines rely erroneously on data in CDF.
I started from ~250 failing tests and got down to zero. But I had to run the full test suite too often which took to long.
It would have been useful to have a flag which allows to rerun only the failed tests. ctest also has that 1 and that is quite useful.