Cleanup containers

[JustinDrake]

1. Use custom types in container definitions (fix Use custom types for container definitions #1153)
2. Renamings

• deposit_index => eth1_deposit_index (more precise and future proof; consistent with eth1_data and eth1_data_votes in "Eth1" grouping)
• previous_epoch_attestations => previous_attestations (consistent with previous_crosslinks, previous_justified_epoch and previous_justified_root; shorter)
• validator_registry => validators (cleaner; more semantically correct—the registry is now both validators and balances; shorter)
• Remove most latest_ prefixes (cleaner; shorter; the prefix does not add much other than clutter; we make an exception for latest_block_header)

3. Add description to not-immediately-obvious container fields
4. Semantically group BeaconState fields, following a logical progression (see item 14 in Phase 0 state transition checklist—take 5 #1054)

    # Versioning
    genesis_time: uint64
    slot: Slot
    fork: Fork
    # History
    latest_block_header: BeaconBlockHeader
    block_roots: Vector[Hash, SLOTS_PER_HISTORICAL_ROOT]
    state_roots: Vector[Hash, SLOTS_PER_HISTORICAL_ROOT]
    historical_roots: List[Hash]
    # Eth1
    eth1_data: Eth1Data
    eth1_data_votes: List[Eth1Data]
    eth1_deposit_index: uint64
    # Registry
    validators: List[Validator]
    balances: List[Gwei]
    # Shuffling
    start_shard: Shard
    randao_mixes: Vector[Hash, RANDAO_MIXES_LENGTH]
    active_index_roots: Vector[Hash, ACTIVE_INDEX_ROOTS_LENGTH]
    # Slashings
    slashed_balances: Vector[Gwei, SLASHED_EXIT_LENGTH]
    # Attestations
    previous_epoch_attestations: List[PendingAttestation]
    current_epoch_attestations: List[PendingAttestation]
    # Crosslinks
    previous_crosslinks: Vector[Crosslink, SHARD_COUNT]
    current_crosslinks: Vector[Crosslink, SHARD_COUNT]
    # Justification
    previous_justified_epoch: Epoch
    previous_justified_root: Hash
    current_justified_epoch: Epoch
    current_justified_root: Hash
    justification_bitfield: uint64
    # Finality
    finalized_epoch: Epoch
    finalized_root: Hash

[djrtwo]

validator_registry => validators

We debated this long ago and decided that validator_registry was more explicit -- it is the entire registry of validators rather than just a list of validators

[djrtwo]

previous_epoch_attestations => previous_attestations (consistent with previous_crosslinks, previous_justified_epoch and previous_justified_root; shorter)

There is a difference between previous_epoch_attestations and previous_crosslinks/previous_justified_epoch.

previous_epoch_attestations are the attestations for the previous epoch. Not the attestations collected during such an epoch.

Whereas previous_crosslinks were the crosslinks during the previous epoch. Same with previous_justified_epoch -- this was the latest justified_epoch during the previous epoch.

[JustinDrake]

We debated this long ago and decided that validator_registry was more explicit

I was in favour of validator_registry in that debate. Since then the situation has changed with the registry split being into two parts:

    # Registry
    validators: List[Validator]
    balances: List[Gwei]

[djrtwo]

Not sure about the semantics here.
The latest_block_header seen should be the parent of the next block, but it is not a parent when recorded and not during any skipped slots when it is recorded into block_roots.

[JustinDrake]

Agreed that naming is awkward here. Brainstorming:

• block_header (matches the type nicely; vagueness hides difficult-to-explain subtleties)
• partial_block_header (a bit more accurate)
• latest_block_header (make an exception here for "latest" prefix)
• latest_partial_block_header (a bit of mouthful)

[djrtwo]

latest_ seems reasonable and purposefully descriptive here (as opposed to it just being thrown on all the history arrays).

[djrtwo]

bump @JustinDrake

Just handled merge conflict. let's resolve this and get merged.

[JustinDrake]

@protolambda, @hwwhww Looks like there is a race condition in the linter CI. I didn't change phase 1 yet the CI is complaining about it.

@djrtwo I think I've addressed all issues, and I've added a bunch of descriptions to field containers. Feel free to edit/add/remove those as you feel appropriate :)

[protolambda]

@JustinDrake I'll take a closer look. Best guess would be that you changed something in phase 0, and it wound up as part of phase 1.
Edit: yes, matching errors in phase-0 and phase-1 spec. Not a race condition. Just your code being phase-1 compatible, with the linting issues included...

[protolambda]

@JustinDrake fixed linting in 75b4692 . Not a race condition, just one missing space. You can run make -B pyspec and make lint to check your changes locally. 👍