Revert raise-on-non-2xx response behavior #54
Conversation
The 0.2.0 release did not include this behavior, so we're reverting it. We may choose to add this behavior back in under a configurable flag in the near future.
spec/oktakit_spec.rb
Outdated
| end | ||
| end | ||
| end | ||
| # describe 'errors' do |
There was a problem hiding this comment.
Let's change the tests to describe the actual behaviour, instead of commenting them out. Since this behaviour is already published and used, we can't safely change it now, so I'd say the tests should describe the actual instead of desired behaviour.
I'd probably treat removing RaiseError as a breaking change, so this would go in oktakit 1.0 or something.
There was a problem hiding this comment.
👍 I agree that removing the raising feature completely would be a major version bump. This next minor version will just be in a weird hybrid state that should work for clients regardless of their faraday version.
Could you expand on this? I don't see anything in the CHANGELOG or release notes that would indicate a change in interface? Since this file exists, it seems clear that the intention was to raise an Oktakit-based error that could later be rescued but now this middleware isn't called at all. It's also worth noting that this change occurred when Faraday was updated and not based on a code change in this repo. |
3b02b32 aimed to fix the originally intended behavior of raising custom errors for most non-2xx responses. This however resulted in a fairly significant change to the actual interface of the 0.2.0 release (in which this behavior seems to have not been working as intended), so we're reverting it.
We may restore the error raising behavior behind a feature flag in an upcoming release.