Which way is better for test data preparation in robot framework?

Question

I’m using robot framework and selenium library to testing a web application, which way is better for test data preparation?

1. Writing test data directly into the test cases, test data act as the user keywords arguments. This way is simple and without any test data file needed, but some user keywords have a bit more arguments and test cases looks strange for people not familiar with.

2. Prepare test data file per test case, then load test data file into variables when executing. This way removes the user keywords arguments and easier for making higher level user keywords, but can’t identify where the variables in the user keywords come from directly and need to open and check the test data file when editing test data.

Answer 1

There is no way best way in general, it will depend on the context (how many tests, how many keywords, how many arguments etc.). Writing Robot tests is like writing code in any other langage: you have to refactor it again and again as it grows.

Though in the specific case of Robot, I agree there is a tension between having short/readable keywords with few/no arguments (solution 1) and more detailed keywords with more arguments (solution 2). My strategy is usually to keep the most important/relevant arguments (like 1 or 2) clearly provided in the test itself and take the other ones from data/lib files. This way you can see what this test is specifically doing without having to check other files.

Answer 2

Best approach depends on the volume and verity of data, if the expectation is to get huge amounts of data to be crunched then, document db i.e mongo db is extermely ood, on all other cases, excel should be good enough as well.