0

Reading through the ink source code (here), my understanding is that a contract call is the selector concatenated with any arguments to the function.

From a Substrate runtime module test, I use the CallData builder which does as described

let mut call = CallData::new( Selector::from_str("foo") );
call.push_arg(&0);

And then make the call through the contracts frame module

let res = <contracts::Module<Test>>::bare_call(      
    ALICE,                                                      
    contract_addr,                                                       
    0,                                                          
    100_000,                                                                                    
    //codec::Encode::encode(&call.to_bytes().to_vec()));        
    call.to_bytes().to_vec());

To a contract function signature of

fn foo(&self, x: AccountId)

This returns a successful execution (Ok(())) but does not invoke the contract function.

It's worth saying that the test is using type AccountId = u64 and ink uses a struct AccountId([u8;32]), but I've tried passing as 32 bytes and it hasn't made a difference.

I've verified that the surrounding test code should work by calling a wasm program written in wat here, which is able to call the exported "call" without passing any parameters.

The complete repositories are also on github for the module and the contract.

Michael
  • 41,989
  • 11
  • 82
  • 128
jay
  • 477
  • 1
  • 9
  • 17
  • This sounds like the contract doesn't actually gets called. If it actually was called and the input data is incorrect then you should see a trap, coming from ink! – pepyakin Jan 08 '20 at 19:02
  • How could it not be called though? Only the contract address could be wrong and that is generated straight from the contracts module https://github.com/jaybutera/nft-registry/blob/master/runtime/src/nftregistry/tests.rs#L394 – jay Jan 12 '20 at 14:01

0 Answers0