Contract deploy contract #439
Conversation
… `ContractError`
d-sonuga
force-pushed
the
contract-deploy-contract
branch
from
595555c to
d107486
Compare
miloszm
left a comment
miloszm
left a comment
There was a problem hiding this comment.
I have some questions about:
- init function call logic
- new struct in the session
| init_arg: Option<Vec<u8>>, | ||
| owner: Vec<u8>, | ||
| gas_limit: u64, | ||
| call_init: bool, |
There was a problem hiding this comment.
why is the call_init mechanism introduced? I think init function needs to be called always if it is present, so far we have assumed that it is so: init called during deployment if present, there should not be a conditionality here
| engine: Engine, | ||
| inner: &'static mut SessionInner, | ||
| original: bool, | ||
| config: SessionConfig, |
There was a problem hiding this comment.
why is the SessionConfig field added here - I think when contract is deploying another contract via an abi, this abi should pass gas limit and price rather than having a session keeping it
There was a problem hiding this comment.
It's there to calculate the deploy charge:
let deploy_charge = max(
bytecode_len * env.gas_per_deploy_byte(),
env.min_deploy_points(),
);| .to_vec(), | ||
| ) | ||
| }; | ||
| env.deploy_raw( |
There was a problem hiding this comment.
why is this not calling init and init is called separately here below?
There was a problem hiding this comment.
The call_inner function that deploy_raw uses seemed to be meant for top-level one-off calls.
Upon completion, the call tree and all instances are cleared. If an init function fails, we don't want the instance that is doing the deployment to be removed.
Aside from that, the memory snapshotting and the applying are things I don't think we need here.
It seemed clearer to me to just call init separately and only do the things that are needed.
For #160.