Can't use XGBoost and TensorFlow at the same time. #728
Description
It seems that executing the train/predict SQL in the Notebook would cause an error, and can not use XGBoost and Tensroflow at the same time.
If using the default sqlflow_submitter which is suitable with Tensorflow, the behavior of using XGBoost will cause sqlflow to throw exception like
sqlflow_submitter.xgboost.common.XGBoostError: XGBoost training task failed: TProtocolException(type=4)
StatusCode.UNKNOWN
training failed exit status 1
This problem can be solved by setting environment variable SQLFLOW_submitter=xgboost. However, this setting is in conflict with TensorFlow.
Is there any way to solve this problem?