Steps to reproduce

1. Have a BigQuery table with the following field (make sure the type=NUMERIC):
   Field{name=risk_level, type=NUMERIC, mode=null, description=null, policyTags=null, maxLength=null, scale=null, precision=null}

2. Use boilerplate code to set up JsonStreamWriter.

3. Call append with an object that has the field above and provide a double. For example: [{"risk_level":2.3}]

Code example

        try {
            JSONObject row = new JSONObject().append("risk_level", 2.3);
            JSONArray batch = new JSONArray();
            batch.put(row);

            boilerplateWriteToTable(batch, "project_name");
        } catch(Exceptions.AppendSerializtionError e) {
            System.out.println(e.getRowIndexToErrorMessage());
            throw e;
        }

Stack trace

getRowIndexToErrorMessage Result:

{0=Field root.risk_level failed to convert to NUMERIC. Error: ByteString scale cannot exceed 9: 2.29999999999999982236431605997495353221893310546875}