NodeId, BackendNodeId, and RemoteObjectId

[brendyna]

Hi @pavelfeldman, @RByers,

We're planning for our next round of EDP implementations. We're digging into DOM/DOMDebugger and we've noticed something we want to get your input on.

There are three types of IDs which we're unclear about the difference between: NodeId, BackendNodeId, and RemoteObjectId.

Could you clarify how those are used within CDP/Chrome DevTools?

Cheers, Brendyn

[robpaveza]

Thanks @brendyna ! I get the difference between RemoteObjectId and the other two, but I don't get the difference between NodeId and BackendNodeId. That's what I'm really trying to figure out.

[paulirish]

cc @dgozman while pavel is OOO

[dgozman]

RemoteObjectId is from a JavaScript world, and corresponds to any JavaScript object, including JS wrappers for DOM nodes. There is a way to convert between node ids and remote object ids (DOM.requestNode and DOM.resolveNode).

NodeId is used for inspection, when backend tracks the node and sends updates to the frontend. If you somehow got NodeId over protocol, backend should have pushed to the frontend all of it's ancestors up to the Document node via DOM.setChildNodes. After that, frontend is always kept up-to-date about anything happening to the node.

BackendNodeId is just a unique identifier for a node. Obtaining it does not send any updates, for example, the node may be destroyed without any notification. This is a way to keep a reference to the Node, when you don't necessarily want to keep track of it. One example would be linking to the node from performance data (e.g. relayout root node). BackendNodeId may be either resolved to inspected node (DOM.pushNodesByBackendIdsToFrontend) or described in more details (DOM.describeNode).

Does this make sense?

[ckpocket99]

Thank you for your explanation of this kind of problem. I'm studying hard