Open ciaranmcnulty opened 3 years ago
which test data are you talking about ?
If you mean files in https://github.com/Behat/Gherkin/tree/master/tests/Behat/Gherkin/Fixtures/features, I think a better way would be to identify cases that are not covered by the cucumber tests but are covered by our fixtures, and to upstream those to cucumber (this way, cucumber parsers would also run against them).
And we would remove any duplicate (which means that once all our upstreaming is accepted, this folder would be empty)
I mean load the Yaml etalon files, run the Go binary against the equivalent features, then see where they diverge.
This could indeed lead to moving them upstream 👍 I think short term having the tests locally will make sense as we can do that immediately.
We'll need to see what the cucumber team's feelings are towards having lots more examples
Now we're running the cucumber testate against Behat, we should also run the Behat test data against Cucumber.
The loader written for those tests should be reusable, the only downside is it'll require the gherkin go binary to execute so maybe should be in an off-by-default set of tests