roadmap
- 01
- 02
- 03
- 04
- 05
- 06
- 07
- 08
Prepare for Cucumber interviews on how Gherkin scenarios become running tests, from step matching to hooks and shared state. Covers BDD practice, the JVM runner, reporting and scenario design.
72 of 72 questions in your plan: 0 known, 0 in progress, 0 due, 0 skipped, 72 new. 0 cut, outside your plan.
Cucumber is rarely a round of its own. It surfaces in QA automation and SDET loops, most often in the framework conversation where you walk through a suite you have built, sometimes as a review of a feature file and its step definitions, and now and then as a live task: turn a user story into a scenario and bind it to code. Developer loops at teams that practise BDD ask about it from the other side, as the outer acceptance test that stays red while the unit tests are written.
The probes are concrete. What does the runner do with a step that two definitions match, and with one that none matches? Which tag expression runs the smoke suite without the work in progress? Why does a scenario pass alone and fail once the suite runs in parallel, when a step class keeps its state in a static field? Where does Cucumber-JVM look for glue code, and which configuration source wins when two of them disagree?
Knowing the keywords, a regular expression and an annotation gets a candidate through the opening questions. What interviewers listen for after that is whether you can name the cost of what was written: steps that describe clicks instead of behaviour, a Background that hides the context a reader needs, a step library that grows one definition per sentence. At senior level, expect to argue whether the Gherkin layer still earns its keep when no one outside the team reads the features.
Cucumber is a tool for a practice, so the plan opens on the practice. Discovery with the three amigos produces the examples, Given-When-Then gives them a shape a business reader can check, and the double loop places the resulting scenario around the unit tests. Those questions ask why a team writes scenarios at all, and they come before any tooling because the tooling only makes sense as an answer to them.
The file comes next. Gherkin's grammar is what the parser reads, and outlines and data tables are how one scenario covers many rows or passes structured input to a step. With the text in place, the plan turns to the code behind it. Step definitions bind each line through Cucumber Expressions or regular expressions, hooks and tags add setup and select what runs, and shared state decides how separate step classes see one scenario's data without leaking it into the next.
Running comes after binding. The Cucumber-JVM runner on the JUnit Platform or TestNG is where glue paths, plugins and parallel execution are configured, and reporting is what that run hands back to people who never open the code. The plan then leaves the JVM for cucumber-js, Behave and Reqnroll, where the same Gherkin binds to a different stack. It closes on scenario design, placed last because judging a step as too imperative or too broad needs every earlier layer in view.
Testers who already write feature files but have never owned the glue code should start at the step definitions and state sharing, where most of the follow-up questions in a framework walkthrough land. Java developers new to BDD should read the practice group first; without it, Cucumber looks like a verbose way to call JUnit, and interviewers notice. With only a few days, narrow level to the range on the job ad and narrow importance to must know alone, then come back for the closing scenario-design module if the role is senior. The plan stops at the step body. Driving a browser from a step, locators, waits and page objects, belongs to the [Selenium roadmap](/roadmaps/selenium). For the whole interview a test-automation role on the JVM runs, with Java, unit testing frameworks and build tools around Cucumber, the [Java SDET roadmap](/roadmaps/java-sdet) carries this tool among the rest.
Testers who already write feature files but have never owned the glue code should start at the step definitions and state sharing, where most of the follow-up questions in a framework walkthrough land. Java developers new to BDD should read the practice group first; without it, Cucumber looks like a verbose way to call JUnit, and interviewers notice. With only a few days, narrow level to the range on the job ad and narrow importance to must know alone, then come back for the closing scenario-design module if the role is senior.
The plan stops at the step body. Driving a browser from a step, locators, waits and page objects, belongs to the Selenium roadmap. For the whole interview a test-automation role on the JVM runs, with Java, unit testing frameworks and build tools around Cucumber, the Java SDET roadmap carries this tool among the rest.