Hi,

Basically you will need to consult the makers of your client application, or the creators of the SDK you would be using to build an application if that is the case.

The NumericRange Matti mentioned is specified in the Write ServiceRequest: https://reference.opcfoundation.org/v104/Core/docs/Part4/5.10.4/#5.10.4.2 (the name is indexRange and it is of type NumericRange), which then points to https://reference.opcfoundation.org/v104/Core/docs/Part4/7.22/, which displays examples of some syntax.

There should be some way to define that, or it is also possible, that your client implementation doesn’t support that at all. Note that even if you write a single element, with that, the value must still be an array (of a single element), not a scalar.

In our Prosys OPC UA SDK for Java, the equivalent parameter would of type com.prosysopc.ua.stack.utils.NumericRange, that has constructors for a single element, single dimension range or multidimensional range.

I would assume most client applications to have some kind of way to define the index range, though it should be mentioned that e.g. our https://www.prosysopc.com/products/opc-ua-browser/ yet cannot write array values, I would recommend to use https://www.unified-automation.com/products/development-tools/uaexpert.html instead if you need that (though, I fail to find an NumericRange setting there either.., maybe it is not so common in actual applications).

I should probably also mention, that there exist some servers (I think I have only seen this in some Siemens’ servers) that seem to make a node per array element, and if the NodeId of the main node is of String IdType (there other types as well of NodeIds), then the subnodes had a square brackets naming scheme that matches typical programming languages array accessors syntax. That is not something specified in OPC UA, but a server implementation is technically free to do so (the e.g. “[0]” etc. is just a set of characters in this case as part of a String, not any operator), so I would not recommend to use it, but if your server happened to be like that, it would be possible to use it. And as an second confusing clarification, there exists a standard modelling rule, ExposesItsArray, that defines that the elements exists also as single nodes, but the specification specifically mentions, that any naming logic is defined on a type level individually, and thus nothing that can really be used in general level implementations.