Change the expectation for predicate methods

[pirj]

Predicate methods typically return true or false, and should only exist for the uncountable and discrete, e.g. use_transactional_fixtures?, but not max_formatted_output_length?.

This relates to the changes made in rspec/rspec-core#2736

[pirj]

I guess those two PRs will be red until any one of those PRs is merged. Unfortunately, our build system is not sophisticated enough to check out the "same branch" to test multi-repo changes.

[pirj]

1. We seem to miss include_examples "adds setting" for use_active_record and file_fixture_path
2. Weird, what fixture_path?/file_fixture_path? are for? Maybe drop them?

Should I fix it here?

[benoittgt]

Thanks @pirj

For 1, I think it could be addressed in another PR
For 2, where do you see fixture_path?/file_fixture_path??

[pirj]

For 2, where do you see fixture_path?/file_fixture_path??

I don't see it used anywhere as well, but if you add this here:

expect(group).to respond_to(:fixture_path?)

it passes, same with file_fixture_path?.

[JonRowe]

rspec-rails overrides rspec-core configuration, so this could be made to work, the question is should it.

[JonRowe]

@pirj I've taken this as base and improved it. Assuming it goes green we should merge it.

My reasoning here:

• The predicates already exist so can't be removed until rspec-rails 5
• We shouldn't depend on a specific version of rspec-core, so just fix it ourselves.

[pirj]

Looks good! 👍

However, regarding backwards-compatibility. We're changing what those predicates return, don't we? Previously file_fixture_path? was returning a string, but now it's a true/false.

[JonRowe]

However, regarding backwards-compatibility. We're changing what those predicates return, don't we? Previously file_fixture_path? was returning a string, but now it's a true/false.

Yes which is my concern with rspec/rspec-core#2736 too but I've come to the decision this can be safely considered a bug fix, its not expected that predicate methods would be used except as truthy/falsely

[JonRowe]

This is green on CI, just hasn't updated Github.

[pirj]

previous definition of use_transactional_fixtures? was here in rspec-core builds. We'll have to resort to attr_accessor.