feat(Simulate): Allow calling view on constrained functions

[spalladino]

Because why not?

This lets an app developer expose a constrained private getter that can be used both for querying state from other contracts and for querying from off-chain by the app, rather than requiring two versions of the same function.

However, implementing this is not as easy as removing the client check. If the function called modifies state, we need to do so in a transient copy of the state, so we can discard it as soon as the call finishes.

[sirasistant]

There are some moving parts for this:

• This would be basically a wrapper of simulate that returns the first non-account contract return values. All changes made by the private simulator are already encapsulated inside the resulting TX, no changes to the DBs are commited on simulate!
• The return values themselves have no form, they are an array of 4 fields. On unconstrained functions, we retain the shape of the return values because the function does not need to return the PrivateCircuitPublicInputs, so we have the shape of the return values in the ABI. For this to be useful, we'd need need some features first:
  • Update the aztec macros to expose the shape of the original return values (before changing them to the public inputs) in the ABI so it reaches the contract artifact
  • Update the return values to be a return values hash similar to the args hash, to have them be arbitrary length
  • In the previously created wrapper, unpack the return values hash and return it to the expected shape by reading the data added to the artifact

[rahul-kothari]

weird question but what is the difference between view() and simulate() - thinking if this might be confusing to the users since we have unconstrained already

[rahul-kothari]

This lets an app developer expose a constrained private getter that can be used both for querying state from other contracts and for querying from off-chain by the app, rather than requiring two versions of the same function.

@spalladino why would you need to duplicate code? A constrained method can call an unconstrained currently (like a normal fn call) right?

[spalladino]

why would you need to duplicate code? A constrained method can call an unconstrained currently (like a normal fn call) right?

The way I see it, it's because you want a constrain on the returned results, which you don't get when calling an unconstrained function. So if you have an app contract that, as part of a tx, calls an unconstrained getter in another contract, the sender can craft a proof where that getter returns pretty much anything.

[nventuro]

weird question but what is the difference between view() and simulate() - thinking if this might be confusing to the users since we have unconstrained already

This is indeed being quite confusing to me as a new user - I expected to be able to call view on any function and get its return value, the same way the Ethereum eth_call RPC works.

Instead I can apparently only call simulate, which returns the very complex Tx object, and it doesn't seem like the data I want is included there (which makes sense due to the transformation into public inputs @sirasistant mentioned).

[LHerskind]

Was made possible in #5432, look at #5762 for decoded return values.