Reach application is dangerous to run on Algorand because of remote call

[lorenzopolidori]

From a Reach application deployed on Algorand I am trying to call an API implemented in a second Reach application using remote:

// ...
const SPP = {
    increaseCapacity: Fun([UInt], UInt)
};

init();

A.only(() => {
    const [tok, price, contractInfo] = declassify([interact.tok, interact.price, interact.contractInfo]);
});
A.publish(tok, price, contractInfo);
commit();

const spp = remote(contractInfo, SPP);
const cap = spp.increaseCapacity(20)

commit()

// ...

When I compile the program, I get the following warning:

WARNING: Compiler instructed to emit for Algorand, but the conservative analysis found these potential problems:
 * Step 1 calls a remote object at /app/blockchain/token-market/index.rsh:46:37:application. This means that Reach's conservative analysis of resource utilization and fees is incorrect, because we cannot take into account the needs of the remote object. Furthermore, the remote object may require special transaction parameters which are not expressed in the Reach API or the Algorand ABI standards.

When I run the programme, I get the following message:

This Reach application is dangerous to run on Algorand for the following reasons:
 * Step 1 calls a remote object at {...}:application. This means that Reach's conservative analysis of resource utilization and fees is incorrect, because we cannot take into account the needs of the remote object. Furthermore, the remote object may require special transaction parameters which are not expressed in the Reach API or the Algorand ABI standards.

and eventually the remote call fails.

Does that mean that effectively we cannot call an API implemented in a separate contract using remote in a Reach application running on Algorand?

[jeapostrophe]

No, it does not mean that. It just means that you need to make sure that it properly. The warning is saying "Reach can't check to make sure you are calling this correctly, so it will fail if you don't". But, Reach does give you the tools to call it correctly, provided you know how. You basically need to extend the call with these things --- https://docs.reach.sh/rsh/consensus/#p_72 --- and if you show the definition of increaseCapacity in the CALLED contract, I could tell you just what you need if you can't figure it out. (You can also compile the CALLED contract with REACH_DEBUG=Y and it will print out basically everything you need to know.)

[jeapostrophe]

Okay. That program doesn't appear to do anything weird. What error do you get? You said "eventually the remote call fails"... what is the error message?

One thing I notice, looking at it again, btw. You are CALLING the function increaseCapacity but the CALLED program has a function named SolarPowerPlant_increaseCapacity.

[lorenzopolidori]

Replacing increaseCapacity with SolarPowerPlant_increaseCapacity fixed the issue, and the call now succeeds.

I still see the warning message multiple times when running the application, is it expected?

This Reach application is dangerous to run on Algorand for the following reasons:
 * Step 1 calls a remote object at /app/blockchain/token-market/index.rsh:46:53:application. This means that Reach's conservative analysis of resource utilization and fees is incorrect, because we cannot take into account the needs of the remote object. Furthermore, the remote object may require special transaction parameters which are not expressed in the Reach API or the Algorand ABI standards.

[jeapostrophe]

Yes, it is expected, because we don't know that the other options aren't needed

[lorenzopolidori]

I see. Do you have any plans to implement something to help Reach understand that the remote call is safe to make, provided certain conditions are met?

[jeapostrophe]

Yup. We have a plan for that.