Running tests against multiple versions of Python seems to make the US Census geocoder more likely to return errors, as discovered in https://github.com/azavea/python-omgeo/pull/64 , possibly due to a high number of requests arriving in a short period. We should find a way to be able to test multiple Python versions without an increased rate of failures.
Running tests against multiple versions of Python seems to make the US Census geocoder more likely to return errors, as discovered in https://github.com/azavea/python-omgeo/pull/64 , possibly due to a high number of requests arriving in a short period. We should find a way to be able to test multiple Python versions without an increased rate of failures.