More tests and checks confirming that ledger.controller is always correct.
#2599
Conversation
|
bot fmt |
|
@gpestana https://gitlab.parity.io/parity/mirrors/polkadot-sdk/-/jobs/4590979 was started for your command Comment |
|
@gpestana Command |
gpestana
added
T2-pallets
| // `ledger.controller` is never stored in raw storage. | ||
| let raw = Ledger::<T>::get(stash).unwrap_or_else(|| { | ||
| Ledger::<T>::get(ctrl.clone()) | ||
| .expect("try_check: bonded stash/ctrl does not have an associated ledger") |
There was a problem hiding this comment.
Should we just log an error or panic? Since try state checks are mostly caused by older issues, we may not want CI to fail for these errors?
Co-authored-by: Kian Paimani <5588131+kianenigma@users.noreply.github.com>
…orrect. (paritytech#2599) A bonded ledger fetched with the `StakingLedger` implementation exposes a method `ledger.controller()` that returns the controller of the ledger. However, that controller is computed and stored under the `ledger.controller` field on the fly - i.e when the ledger is fetched from storage using the `StakingLedger::get` method. The controller field is never stored in storage. This PR add a few more tests checks and improves the ledger try-state checks to make sure these invariants hold true. --------- Co-authored-by: command-bot <> Co-authored-by: Bastian Köcher <git@kchr.de> Co-authored-by: Kian Paimani <5588131+kianenigma@users.noreply.github.com>
A bonded ledger fetched with the
StakingLedgerimplementation exposes a methodledger.controller()that returns the controller of the ledger. However, that controller is computed and stored under theledger.controllerfield on the fly - i.e when the ledger is fetched from storage using theStakingLedger::getmethod. The controller field is never stored in storage.This PR add a few more tests checks and improves the ledger try-state checks to make sure these invariants hold true.