Behaviour of get_node

[Randelung]

Hi, I was wondering about the behaviour of get_node and its stability.
Right now you can get_node any arbitrary ID and get a Node object, without any checks whether the node exists or not. You only discover that it doesn't when you try to interact with it.
I've inadvertently used that mechanic in the setup of some data binding in my program. The nodes are being geted before they exist on the server and a callback that writes to the node is set up.
After, the nodes are actually being created, and it's only then that the nodes will be interacted with.

I'm trying to gauge if that might change in the future and possibly break that setup, as it's possible you might include a check.
Do you have any plans of modifying this behaviour?

Thanks!

[AndreasHeine]

actually the node class is just a gathering of usefull highlevel methods otherwise you need to make lowlevel calls (like in uaclient). usually in my clients i try to avoid the nodeclass as much as i can because its one request for just one part of information (value, browsename, ...). i prefer to bundle those request and read multiple nodes attributes what so ever in just one request to reduce the network traffic (but check the servers maxnodesperbrowse, etc up front^^)! if you trace your opc ua application you will see a lot of unnecessary requests (at least my oppinion)

a checkup from serverside wouldnt be so costly but from client side... it would be a request to verify the node exist.

[Randelung]

OK cool, thanks!

[jmarshall9120]

@AndreasHeine - Could perhaps provide and example on how your query multiple nodes in one request? I haven't found any examples of that.

[AndreasHeine]

@AndreasHeine - Could perhaps provide and example on how your query multiple nodes in one request? I haven't found any examples of that.

 https://github.com/AndreasHeine/opcua-tutorial/blob/da44b65682ff70ace7feb640d6d43c7d2880f7ed/client/read.py#L22

[jmarshall9120]

@AndreasHeine

That doc you linked is an awesome resource! One question though, are we supposed to assume that the results are synchronous when using client.uaclient.read_attributes?

I noticed the result object (I was testing with DataValue results by passing ua.AttributeIds.Value) doesn't return any information linking it to the NodeId I passed. Due to this I would have to assume the data returned in the same order as the list of NodeId passed to client.ua.read_attributes. This is an assumption one would never make on something like an SQL server.

[AndreasHeine]

opc ua spec. says that it has to be in the same order like the requested nodeids so the result idx matches