Release 6.7.0

Features

• Write Cucumber JSON reports and verbose output with Swift Testing (#420, #433 by @ianhlavats)

• Run feature hooks and attachments with Swift Testing (#421, #435 by @ianhlavats)

• Bring the Swift Testing runner's step and hook options in line with CucumberSwift (#422, #434 by @ianhlavats)

• Run Swift Testing scenarios in parallel (#432, #436 by @ianhlavats)

• Add expectations, readable test names and a test per step to the Swift Testing runner (#441, #444 by @ianhlavats)

• Add macro forms of the hooks (#443, #445 by @ianhlavats)

• Suggest missing step definitions as macros (#453, #456 by @ianhlavats)

• Make the Swift Testing JSON report match the XCTest report's uri and file name (#463, #466 by @ianhlavats)

  Migration:

  In Swift Testing mode, Feature.uri, as hooks, shouldRunWith(scenario:tags:) and custom observers see it, is now the file:// URL of the feature file's copy in the test bundle, as with XCTest, not a plain source path. Code that builds a URL with URL(fileURLWithPath: feature.uri) should use URL(string: feature.uri) instead. Failures in Xcode still point at the editable feature file. CucumberTestResultsFor<target>.json is now written in Swift Testing mode too, and in both modes it is replaced on every run instead of keeping the first run's report.

• Let a step skip the rest of its scenario in Swift Testing, as XCTSkip does in XCTest (#470, #471 by @ianhlavats)

Bugs

• Fix wait(for:timeout:), which never fails a step (#442, #449 by @ianhlavats)

  Migration:

  wait(for:timeout:) now fails the step, with XCTest and with Swift Testing, when the timeout passes before every expectation is fulfilled, or when an inverted expectation is fulfilled, as XCTest's wait(for:timeout:) fails a test. Until now such a wait passed, so a scenario whose wait times out will now fail. To fix it, make sure the expectation is fulfilled within the timeout, or raise the timeout. In an async step, write await wait(for:timeout:): a synchronous wait there holds the main actor, so work on the main actor or the main queue can't fulfill the expectation.

• A Rule's Scenario Outlines never run, and its tags are dropped (#468, #469 by @ianhlavats)

  Migration:

  Scenario Outlines inside a Rule used to be skipped silently. They now run, with the feature's and the rule's Background steps, so a suite can see new passing or failing examples. Tags above a Rule used to reach only the rule's first scenario; they now apply to every scenario and example in the rule, after the feature's tags and before each scenario's own. A tag filter (CUCUMBER_TAGS or shouldRunWith(scenario:tags:)) that selects a rule's tag now selects all of the rule's scenarios. Check filters that relied on the old behaviour.

Tasks

• Parallel tests (iOS, UI, Xcode 16) often fail because the UI test can't start or stop the fixture app (#375, #431 by @ianhlavats)

• Document how to use the step definition macros (#419, #430 by @ianhlavats)

• Separate what the Swift Testing runner leaves out by design from what it lacks (#423, #455 by @ianhlavats)

• Deprecate ExecuteFirstStep in favour of shared helper functions (#446, #447 by @ianhlavats)

  Migration:

  ExecuteFirstStep is deprecated in both runners and will be removed in the next major version. It still works, with a deprecation warning. Move the shared logic into a helper function that both step definitions call (see "Running another step" in Matching Steps).

• Let mise run test-fixtures run only the fixtures named in FIXTURES (#448, #450 by @ianhlavats)

• Run SwiftLint in CI so lint errors can't reach main (#451, #454 by @ianhlavats)

• Show the step and hook macros in every documentation example (#452, #457 by @ianhlavats)

Other changes

• chore(actions): bump the codeql-action group with 2 updates (#460 by dependabot[bot])
• chore(actions): bump actions/create-github-app-token from 3.2.0 to 3.3.0 (#461 by dependabot[bot])

Full list of changes: 6.6.0...6.7.0