Hi,

We can consider that for a future version. In the meantime you could try to do it more manually (and let us know if that is convinient enough). Internally the autoConvert will do following:

1. Read the DataType Attribute of the node to be written. This must be done each time, since on that API level we don’t have it otherwise (UaVariableImpl would know; see below). Thus there is a small performance hit due to that.

2. Convert the given value with (UaClient).getAddressSpace().getDataTypeConverter().convert(Variant, NodeId of the datatype). This includes first a check if the initial Object was DataValue, Variant or raw value (DataValue.getValue, Variant as-is, and new Variant(object) for raw values.

3. Delegate to other non-autoConvert methods.

So you should be able to convert them individually first and then use the writeValues(NodeId[], Object[]). However you will need to read the DataTypes of all the nodes once (assuming they do not change, you could cache that). Alternatively you could get the UaNodes via client.getAddressSpace().getNodes(). Cast the UaNodes to UaVariable(Impl). And then you can .getDataTypeId() to get the DataType (or .getDataType, to pass the UaDataType node directly to DataTypeConverter.convert(Variant value, UaDataType dataType)).

P.S.
While the autoConvert option is easy to use (that’s why it was added), I would say the “more proper” and performant way would be to conversion manually like this as 1) no need to Read the DataType all the time (assuming you can cache them) 2) A bit more separation on conversion errors from write error (the method does throw DataTypeConversionException though, however that will be also thrown e.g. if the DataType Attribute cannot be Read.).

P.S.2.
There is also UaVariableImpl.writeValue(Object value, boolean autoConvert), but that too will write only one value at a time (but that doesn’t need to read the DataType all the time). Maybe at some point we can get a write API to use those instead of NodeIds in bulk.