improve testing

[gonsie]

There are a number of task that should be done to improve the testing:

• improve the setup of the Travis tests. This should involve using multiple build stages and conditional stages to test various compilers and deploy the doxygen.
• improve how we leverage PHOLD to test various features. There should be a distinction between an instrumented phold version which can do all the things and the original phold benchmark (I'm not sure if one model code base can/should capture both of these concepts.)
• through the improvements to a feature-rich phold test, improve our test coverage.
• improve the cmake for generating coverage (refactor for bugs uncovered in see if gcov and clang versions match #150)
• improve the cmake for running tests. Right now mpirun -np is hard-coded, but it doesn't need to be.

[caitlinross]

To add on to this, coverage decreased quite a bit with the merge of #149. I decided to go ahead and merge it anyway and we can work on improving coverage with the rest of these tasks.

[caitlinross]

FYI, next month Kitware is having a CMake tutorial day. Neil and I are both going, so I plan to work work on making some more improvements to our CMake files and testing around then (plus I should be done with my thesis by then).

[laprej]

Any updates on this? Did you guys learn any cool tips and tricks with CMake? :)

[caitlinross]

Not yet. The CMake tutorial is on the 23rd! Maybe they still have some seats open if you want to attend.

[laprej]

Excellent, I registered, thanks!