Why should we use writeFinal when recording a final state to Vstorage

[Jorge-Lopes]

When we retrieve a recorder object from the makeRecorderKit API, we get both write and writeFinal methods, with the difference being that the latest will prevent the publisher from receiving more calls.
recorder.js

// below here differs from writeFinal()
this.state.closed = true;
return publisher.finish(value);

I understand that the best practice is to use writeFinal when we are executing a final call to publish the completion value. Which will prevent that node value from being updated again.

What I would like to understand in more depth is if this pattern was implemented to protect the developer in case of fault design in the contract or if it has any overload to Agoric chain to keep the Publish object listening to updates?

Thank you all in advance

[dckc]

@gibson042 ? thoughts?

[gibson042]

What I would like to understand in more depth is if this pattern was implemented to protect the developer in case of fault design in the contract or if it has any overload to Agoric chain to keep the Publish object listening to updates?

The former. Our async sequence publisher pattern consists of a { publish, finish, fail } trio as described at @agoric/notifier and Notifiers and Subscriptions, and the RecorderKit interface is built on top of it. Also, finalization applies to the publisher object but not to the chain storage node that ultimately receives its data—a distinct publisher object associated with the same chain storage node can still continue writing to it.

Is that sufficiently clear?

[Jorge-Lopes]

Yes it is @gibson042.
Thank you very much for the explanation.