NULL can have several definitions:
The branchId can be a null NodeId (not a general NULL)
The Comment is a localized text field - again it can be a NULL localized text (which has a definition of what a Null is)
The ClientUserId is a string and needs to be a NULL string.

Is this what you were providing?

Yes I believe I am. See attached pictures from UaExpert showing that each field has the correct data type.

BranchIdClientUserIdComment.png (18,750 bytes)   

BranchIdClientUserIdComment.png (18,750 bytes)   

Test_005.png (14,729 bytes)   

Test_005.png (14,729 bytes)   

I see OPC 10000-3 8.2.3 has a special definition for what a Null NodeId is. Setting BranchId to ns=0,i=0 silences the complaint about the BranchId value.

I cannot see anything in the specification that gives a special definition for LocalizedText or string.

On second thoughts, does 10000-3 8.2.3 really preclude a server from returning a null variant NodeId?

It reads to me that if you specify a numeric 0, null or empty string, a zero guid, or a bytestring with length 0 as the NodeId value, then you must set NamespaceIndex to 0. But that doesn't imply that a variant cannot contain null for a NodeId data type.

NodeId is a sub-type of BaseDataType. 10000-3 8.7 states that BaseDataType "defines a special value null indicating that a value is not present".

See also version 1.05 that strongly supports my argument that a variant null should be allowed:

"A canonical null NodeId has an IdType equal to Numeric, a NamespaceIndex equal to 0 and an Identifier equal to 0.

In addition to the canonical null NodeId the alternative values defined in Table 23 shall be considered a null NodeId."

This test is giving an error if the "alternative value" is not provided.

Resolved with Mantis Issue 8315

Agreed the issue is a duplicated - closed issue