Seite - 3 - in The Future of Software Quality Assurance

Change-Driven Testing 3 Code Version History Test Coverage Tickets Test Runtime … Test Intelligence » Wherearewemissing tests? « » Which tests do weneed to run? « » Which tests are redundant? « » What causesa particular test failure? « » …? « Fig. 1 Test intelligence: combining readily available data from various data sources in the software development process to automatically answer questions about testing Thequestionswewant toanswerthroughTest Intelligenceconcernbothourtests and thecodechangesunder test. Therefore,we collectdataaboutboth the tests and thechanges.Muchof thisdata is alreadyavailable in thedevelopmentenvironment andneedonlybeextracted forourpurposes. To communicate the extracted data to testers, developers, and managers alike, we use treemaps. Figure 2a shows such a treemap that represents the code of a UI component of the software systemTeamscale. Each box on the map represents a single method in the code. The size of the box is proportional to the size of that methodin linesofcode.We color theboxes to highlightparticularpropertiesof the respectivecode. 2.1 Version-ControlSystems Version-control systems (VCS), such as git or TFS, are a de facto standard in today’s software development. Generally speaking, a VCS keeps a chronological history of code changes committed by developers and helps them coordinate their changes.Within theVCS,changesmaybeorganizedinbranches,whereabranch is an isolated lineofchanges that isbasedonaparticularversionof thecodeandmay be merged into a later version of the code. If developersuse a dedicated branch for the implementationofa particular feature,we call it a featurebranch. From the change history in a version-control system, we may extract the list of changes since any given baseline, be it a particular release, the last test run, the start of a feature branch,or any other point in time. Figure 2b shows code changes on a treemap. Methods that were added by one of these changes are highlighted in