Getting the full response with the ResponseInvalidError about openapi_first HOT 3 OPEN

By the way, if you use request.validate_response(rack_response) without passing raise_error: true you get an object where you can get all details including the response body. Somehing like:

response = Rack::Response[*app.call(env)]
request = Rack::Request.new(env)
validated = definition.validate_response(request, response)
unless validated.valid?
  validated.body # => parsed response body
  validated.error # => Failure instance
  validated.error.errors.map(&:message)
end

from openapi_first.

Hi. I think it makes sense to add more details to exceptions.
I think it would be best to add ResponseInvalidError#errors or ResponseInvalidError#failure.errors to return an array of error objects (Schema::ValidationError) coming from the schema validation result and use that to get detailed information about what went wrong during validation without leaking too many details of your actual response body.

Making the actual response available on the exception makes sense to me as well for debugging, but I would not use that to log the response body.

from openapi_first.

By the way, if you use request.validate_response(rack_response) without passing raise_error: true you get an object where you can get all details including the response body. Somehing like:

Thanks! That seems to work nice.

Logging at this point is just because some minor errors (mostly missing null being allowed) are still too abundant.

Once i got the bulk going without errors i will switch over to send validation problems to Sentry and later have request and response validation active all the time on our public development server (opposed to just on my local dev machine).

from openapi_first.