Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Commits on May 6, 2015
  1. @mattwynne

    Merge pull request #846 from pdswan/regression/missing-scenario-status

    mattwynne authored
    Expose #status on Cucumber::RunningTestCase [#836]
Commits on May 4, 2015
  1. @mattwynne

    Merge pull request #843 from cucumber/result-new-api-usable

    mattwynne authored
    Make the result arguments of the new formatter api produce the proper backtrace
Commits on May 1, 2015
  1. @pdswan
  2. @pdswan
Commits on Apr 30, 2015
  1. @mattwynne

    Merge branch 'exit-status-codes' of

    mattwynne authored
    … into ciembor-exit-status-codes
  2. @tooky

    Merge pull request #842 from pdswan/find-all-steps

    tooky authored
    Default require path to features unless specified explicitly.
  3. @ciembor

    Exit with different status codes when scenarios failed (1) and when a…

    ciembor authored
    …pplication exited with error (2).
Commits on Apr 25, 2015
  1. @brasmusson

    Failing test for the junit_formatter changes

    brasmusson authored
    Update the junit_formatter to use only use after_test_case for
    Scenario Outlines in expanded mode. Update the junit_formatter
Commits on Apr 24, 2015
  1. @brasmusson

    Document the Junit formatters behaviour for Scenario Outlines

    brasmusson authored
    Update the junit_formatter.feature to cover more cases with respect to
    Scenario Outlines.
  2. @pdswan

    Default require path to features unless specified explicitly.

    pdswan authored pdswan committed
    In general there seem to be some weird dependencies (or lack there of) between
    configuration options and information in `Cucumber::Cli::Configuration#feature_files`
    that could probably be clarified with a little refactoring.
    One example that this feature exposed directly is that
    `Cucumber::Cli::Configuration#feature_files` does not
    depend on `Cucumber::Cli::Configuration#feature_dirs` though their names
    certainly imply dependence.
    I had deleted `Cucumber::Cli::Configuration#feature_dirs` to eliminate
    the confusion, but added it back in when I realized it was used in
Commits on Apr 3, 2015
  1. @brasmusson
Commits on Mar 26, 2015
  1. @mattwynne

    Clean up some features

    mattwynne authored
  2. @mattwynne
Commits on Mar 24, 2015
  1. @tooky
  2. @olleolleolle @tooky

    Tell users to use the --init command

    olleolleolle authored tooky committed
     Inspired by #699.
  3. @tooky

    Reverse the order of after hooks when applied to a test case.

    tooky authored
    Add a feature documenting the order after hooks should be executed.
    cc. @mattwynne
Commits on Mar 19, 2015
  1. @tooky
  2. @tooky

    Nested steps are really steps that are invoked dynamically.

    tooky authored
    Catch undefined steps inside `invoke_dynamic_step` so that we can report
    them as errors to the user.
  3. @tooky
  4. @tooky
  5. @mattwynne @tooky

    Undefined nested steps

    mattwynne authored tooky committed
    This scenario highlights a regression about how we print snippets for
    undefined, nested steps.
    What's the right behaviour here? Previously, we treated these as the
    same as undefined steps from features, but now I've looked into this,
    I'm not so sure.
    This scenario describes the existing behaviour, but is it desirable?
    If you have a nested step, surely you expect it to already match? If
    it fails to match, I think you just want that to be a straight up
    failure, rather than a regular "undefined" error that prints a snippet.
    In the new 2.0 model, we know about undefined Gherkin steps as soon as we've
    compiled the mappings, but we can't know about nested steps until we
    execute the action, so they're different. It feels like the model might
    be telling us something.
  6. @tooky
Commits on Mar 18, 2015
  1. @mattwynne
  2. @mattwynne

    Send `end_scenario` message to programming languages

    mattwynne authored
    In the future, the programming languages should do this themselves
    by adding their own filters, but this is the most direct way to fix
    the regression #813.
    Closes #813
  3. @tooky

    Prepare world as an around hook.

    tooky authored
    This ensures that prepare world is the first thing that is
    run for a test case, preventing the inconsistency with what the world _is_
    in around hooks.
    Legacy adapter requires each test case to have a test step otherwise it
    doesn't switch step container and messages get lost.
    Also ensures failures in Around hooks are reported.
  4. @maxlinc @tooky

    Test showing World is inconsistently available in Around hook

    maxlinc authored tooky committed
Commits on Mar 17, 2015
  1. @brasmusson

    Remove keyword from name of the scenario object used in hooks.

    brasmusson authored
    Fixes #768 (the keyword part).
Commits on Mar 16, 2015
  1. @mattwynne
Commits on Mar 15, 2015
  1. @brasmusson
Commits on Mar 14, 2015
  1. @brasmusson
  2. @brasmusson
Commits on Mar 13, 2015
  1. @mattwynne

    Tidy up wire protocol features

    mattwynne authored
    - use markdown and gherkin description fields
    - split general "wire protocol" feature into one feature per message
    - add a readme overview
  2. @mattwynne
Commits on Mar 4, 2015
  1. @brasmusson

    Pass when using both --dry-run and --strict. Fixes #810.

    brasmusson authored
    Fail only when there are undefined steps when using both the --dry-run
    and the --strict options. Fail when there are pending steps with the
    --strict option.
Commits on Feb 14, 2015
  1. @brasmusson

    Delay step output until after step has been called. Fixed #806.

    brasmusson authored
    Delay sending the step output to the gherkin formatter in the
    JSONFormatter, also for Scenario Outlines in expanded mode, until step
    has been called on the gherkin formatter.
Something went wrong with that request. Please try again.