Last Friday, my mentors advised me to set up Travis configuration so that Python 2.7 virtual machine in Travis would test horton 2 bridge. This was possible because some of other packages that cclib has bridge to are not maintained for Python 2 anyway. This cleared up dependency issues that were present earlier.

Today, I therefore wrote up necessary configuration for Travis to fetch horton 2 from conda. Then, I finished up test function for the bridge. Basically, it loads a sample data file already included in cclib repository (dvb_un_sp.fchk). This is a simple path I can take because both cclib and horton 2 supports loading from Gaussian fchk files. For the makehorton function, I am basically comparing the values between the horton IOData object that was loaded using from_file() function in horton and object that was created from cclib data. The other way around works for makecclib function.

When I am able to clear up all issues regarding this bridge to horton 2, I will be able to work on bridge to horton 3 and a wrapper for Hirshfeld type methods. A different (possibly more sensible) testing may also be one of the objectives I will think about.