[jvm-package] how to make xgboost4j-spark use DenseVector to build model on version 0.90

First, thanks for your great job.
I have one question which need your help.
Our dataset has a lot of 0.0 as meaning value, so we do not want set 0.0 as missing value. What I can do is to make xgboost4j-spark use DenseVector. But I do not know how to do this.
Can anybody help? Some examples would be appriciated.
Thank you.

Start by checking the xgboost4j-spark 0.90 input-vector handling and any existing examples for DenseVector and missing values. Reproduce the dataset setup described in the issue; done means providing a documented, maintainer-confirmed example or clearly recording that the requested input is unsupported.

Written by the indexing model from the issue text.