Add mypy type hinting check #1166
Conversation
| for challenge in state.custody_bit_challenge_records: | ||
| if get_current_epoch(state) > challenge.inclusion_epoch + CUSTODY_RESPONSE_DEADLINE: | ||
| slash_validator(state, challenge.responder_index, challenge.challenger_index) | ||
| for custody_chunk_challenge in state.custody_chunk_challenge_records: |
There was a problem hiding this comment.
Workaround for mypy issue of redefining variable in the different scopes (python/mypy#5750).
|
Lazy benchmarking on running |
|
Re: benchmark |
specs/core/0_beacon-chain.md
Outdated
| @@ -1140,15 +1140,15 @@ def slash_validator(state: BeaconState, | |||
| current_epoch = get_current_epoch(state) | |||
| initiate_validator_exit(state, slashed_index) | |||
| state.validator_registry[slashed_index].slashed = True | |||
| state.validator_registry[slashed_index].withdrawable_epoch = current_epoch + LATEST_SLASHED_EXIT_LENGTH | |||
| state.validator_registry[slashed_index].withdrawable_epoch = Epoch(current_epoch + LATEST_SLASHED_EXIT_LENGTH) | |||
There was a problem hiding this comment.
If current_epoch and LATEST_SLASHED_EXIT_LENGTH are both of type Epoch do we need to wrap the sum with Epoch( ... )?
There was a problem hiding this comment.
Epoch is defined as "EpochNumber/EpochID" and LATEST_SLASHED_EXIT_LENGTH is defined as "diff between Epochs".
I'm slightly in favor of the status quo but I'm open to it; for Epoch and Slot, we can change it to the more flexible definition if we want. :)
|
I'm not so happy with the benchmark results but it's ready to get the first review. There are some other
/cc @JustinDrake what would you say? |
| slot: uint64 | ||
| shard: uint64 | ||
| slot: Slot | ||
| shard: Shard | ||
| beacon_chain_root: Bytes32 | ||
| parent_root: Bytes32 |
There was a problem hiding this comment.
All the hashes can be of type Hash (see PR1156 where I introduce the type).
|
@JustinDrake up-to-date now with #1156 |
X = NewType('X', NONE-BASE-TYPE)(Fine-grained: Inconsistent behavior for nested NewType python/mypy#4615), I changeduint64toclassto make the check pass. I'll benchmark it to see if it increased too much overhead.BytesNcustom types familyTODOs: