The problem is within to_not receive(...).once, which is not a valid way to express an expectation.
On previous releases of rspec it can be attributed quite the opposite behaviour. It would be turned into an expectation of being called exactly once. That would also explain a passing spec before.
Sorry, I saw your second PR and realised that the build was failing, so I went and fixed it before I saw this PR. Closing, as it should be passing now.
I've noticed that specs are failing when run on
'rspec', '= 3.8.0'
(see https://travis-ci.org/webmachine/webmachine-ruby/jobs/465823764).The problem is within
to_not receive(...).once
, which is not a valid way to express an expectation.On previous releases of rspec it can be attributed quite the opposite behaviour. It would be turned into an expectation of being called exactly once. That would also explain a passing spec before.
Related issues from
rspec-mocks
are https://github.com/rspec/rspec-mocks/issues/1210 and https://github.com/rspec/rspec-mocks/pull/1212 and changelog.