The rdataretriever project not only exposed me to systematic coding, but it also introduced me to Testing and Travis. I learned how to deal with Travis environment and its different and weird behaviour when it comes to intrepreting paths. In the beginning, it was little overwhelming because all I could see was “red crosses” on my dashboard with occassional “green” ones (Breathing a sigh of relief) after consistent updates to the config file in the hopes that it will pass.
Continuous integration for packages is a whole new world and very important as it is crucial to know what a single line of updated/changed code could do to your work. I say it as a whole new world, because manual testing is very different. Testing for rdataretriever with reticulate is a myraid scenario due to considerations of different OS, different R platforms, virtual / non-virtual environments. Within these combinations, the interpreration of paths change too.
Therefore, if any error is encountered it has to be checked in system each having a different permutation, which involves cleaning up the system, setting the databases, environments etc. again and then moving on with the testing. Dealing with Travis requires a different perspective and deep understanding about its configuration files and the manner it interprets the commands.
Currently, after mentioning explicitly lines for using testthat, the test cases are passing. But ideally,it is expected of Travis to run the test for particular package without explicitly mentioning it. In this case, it is having a little trouble with finding DBI package. The root of the cause might be different, that is, it might not be occurring just because of interpreation of paths, but resolving it is still in progress.