If a test named “Foo” fail and a test named “FooBar” fails, it looks like both will be rerun because of the failure of foo.
Exactly like what happened if ones run ctest -R Foo, Foo and FooBar will be run.
This should be improved, but I guess it’s on cmake side.
A way to improve that would be to never have a test name included within another test name.
An example of this actually happening :
https://buildbot.kitware.com/builders/paraview%40master-luigi-linux-shared-release%2Bgui%2Bmpi%2Bpython2/builds/1133/steps/build-n-test/logs/stdio
pvcrs.XdmfReadStructuredGrid and pvcrs.XdmfReadStructuredGridCollection