Add afterAll hooks

[jbpros]

From #84 (comment).

[mwinteringham]

+1

[JRedgrave]

+1

[aslakhellesoy]

What are all these strange patches that add a single line with a 1?

I'm not sure that implementation is complete!

[namukang]

+1

[3Thomson14]

+1

[ahoffmann]

+1

[corinna000]

+1

[DavidSouther]

Any progress on this in the past year?

[jbpros]

While this is not a definitive solution, the recently-added registerHandler() method (available on NPM as of 0.3.2) can help you run code at the end of the run:

module.exports = function () {
  this.registerHandler('AfterFeatures', function (event, callback) {
    console.log("finished running scenarios", event);
    // do your stuff
    callback();
  });
};

registerHandler() is considered unstable API at the moment. I'm not sure we'll keep it in and we still want a decent this.AfterAll() helper at some point.

One potential issue I see (and tested) with this solution is that calling back with an error will not make Cucumber fail the run. @devpaul is that a bug in your opinion?

[devpaul]

@jbpros I believe it should be considered a bug. It doesn't follow convention.

[jwoudenberg]

To me, it wouldn't make sense if an event handler can make the piece of code that triggered it throw an error, that's now how I expect events to work (they listen).

Also, the handler does not get the world as a context as reported in #244. Again, from the perspective of it being an event, that makes sense to me. When using it as an AfterAll hook however, this is a serious shortcoming.

[djungowski]

@jbpros This does not help either since a handler registered this way is executed after every feature file.

[djungowski]

Furthermore, the README is completely misleading on the behavior of AfterFeatures:

The after features event is emitted once all features have been executed, just before the process exits. It can be used for tasks such as closing your browser after running automated browser tests with selenium or phantomjs.

No, it is not. It is executed after every set of features (=feature file)

[jbpros]

@djungowski no, AfterFeatures is triggered only once per run, after all features and their scenarios were processed. The README section related to it looks good to me. How do you think it's misleading?

[charlierudolph]

Closing as registerHandler now has full support and the passed function can use any of the available interfaces that steps/hooks use and when passed an error it will throw it and kill the test run.

[jwbuitenhuis]

In which version is that enhancement? I've tried it in 0.10.4 but it doesn't seem to respond to callback(new Error('foo')).

[charlierudolph]

In the future please see the changelog. Searching through it, it appears it was added in 1.1.0

[alayor]

@charlierudolph Could you tell me how I can use registerHandler using defineSupportCode?

var {defineSupportCode} = require('cucumber');

[charlierudolph]

Please see the docs for "Event Handlers". There is a link on the README

[kyleian]

Hate to pull this one out of the grave again, but per @charlierudolph's comment, can you clarify the expectation is that passing an error to the registerHandler callback should infact kill the entire test run?

In my use case/need, I'd like to fail exclusively the scenario if an error is raised during the AfterScenario hook, but continue the suite after marking that scenario as failed (in the way I can during a step); is this achievable with 1.3.1+? My example usecase would be in the event of failing to tear down mocks generated during the scenario.

e: after looking around for some different examples this comment seems to directly answer my question.

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.