Batch- and model-driven test automation
In Chapter 1, we discussed high-volume test automation. In the case study, we had a system with no user interface. What the system did was take from text files to populate a database. When Matt was at the insurance company, he wrote a number of these extract, transform, and load (ETL) programs, sometimes just extract. For example, say an executive wants a list of all the customers who have Diamond status. There is no user interface or “state”, just a data file to look at. The simplest method for independent reporting is likely to have someone else pull the data and compare the answers, combined with a random sampling. Sometimes this approach is called “write it twice.” At the insurance company, for example, there was an independent reporting team, and they might write a second database query to answer the question. The results of the queries could be compared and, as long as the text file matched the results, the team...