refactor(vara.eth): introduce typed MB/EB hashes, move MB to ethexe-common#5508
Draft
grishasobol wants to merge 4 commits into
Draft
Conversation
grishasobol
added
type: refactor
grishasobol
added
the
ai-friendly
grishasobol
force-pushed
the
gsobol/ethexe/malachite-core-concretize-payload
branch
from
733e281 to
eca538f
Compare
grishasobol
force-pushed
the
gsobol/ethexe/malachite-typed-block-hashes
branch
from
2b05f34 to
a5e2859
Compare
grishasobol
added a commit
that referenced
this pull request
May 25, 2026
PR #5508 left MbStorageRO/RW and OnChainStorageRO/RW + BlockMetaStorageRO/RW methods taking raw H256, which forced every caller to thread untyped hashes and use `.inner()` at boundaries. Type the signatures end-to-end so the typed values flow through the codebase: EB-block hashes become `HashOf<EB>`, MB hashes become `HashOf<MB>`. CAS payload hashes (`set_transactions` return, `transactions(payload_hash)`) stay raw H256 since they're content-addressed, not entity-addressed. The cascade is real — ~30 files touched across ethexe/common, ethexe/db, ethexe/compute, ethexe/consensus, ethexe/malachite/service, ethexe/network, ethexe/observer, ethexe/prometheus, ethexe/rpc and ethexe/service. Where a function receives raw H256 from an external source (chain events, malachite-core API, test fixtures) we still cross the typed boundary with a local `mb_h` / `eb_h` helper rather than making `HashOf::new` safe, so each conversion remains explicit and auditable. `Mirror.processOutgoingAction` flow and existing tests continue to compile and lint clean across the workspace.
Member
Author
|
Pushed refactor to type DB trait signatures with CI: https://github.com/gear-tech/gear/actions/runs/26400404637 |
grishasobol
force-pushed
the
gsobol/ethexe/malachite-core-concretize-payload
branch
from
fbd5c67 to
d3341a6
Compare
…d to ethexe-common
Move the consensus block envelope `MB` (formerly `ethexe_malachite_core::Block`),
the versioned `BlockPayload`, and the on-disk `CompactMb` shape into
`ethexe_common::malachite` so the executor (`ethexe-compute`, etc.)
can reason about them without depending on the consensus layer.
Type-safe MB / EB hashes:
- `CompactMb { parent: HashOf<MB>, height, payload_hash, reserved }` —
same shape as `MB` with the opaque payload bytes replaced by the
payload digest. Renames the legacy `transactions_hash` field.
- `MbMeta::last_advanced_eb: HashOf<EB>`.
- `BlockMeta::{last_committed_mb, last_committed_eb}` typed.
- `DBGlobals::{start_block_hash, latest_prepared_eb_hash,
latest_finalized_mb_hash, latest_computed_mb_hash}` typed.
- `DBConfig::genesis_block_hash: HashOf<EB>`.
- New `pub type EB = SimpleBlockData` alias + `SimpleBlockData::typed_hash()`
/ `eb_hash_of(H256)` for lifting raw Ethereum block hashes into the typed
envelope.
`HashOf<T>` SCALE-encodes as a bare 32-byte hash (phantom is `#[codec(skip)]`),
so the on-wire/on-disk byte layout is identical to the previous bare
`H256` columns where these fields used to be plain hashes.
Trait sigs in `MbStorageRO/RW` keep `mb_hash: H256` for now — the gain
from threading `HashOf<MB>` through every storage call site (compute,
cli, consensus, processor, etc.) doesn't justify the cascade right now;
typed wrappers at boundary structs already catch the EB-vs-MB confusion
the refactor was about. Call sites pass `field.inner()` where needed.
Fixes #5507
PR #5508 left MbStorageRO/RW and OnChainStorageRO/RW + BlockMetaStorageRO/RW methods taking raw H256, which forced every caller to thread untyped hashes and use `.inner()` at boundaries. Type the signatures end-to-end so the typed values flow through the codebase: EB-block hashes become `HashOf<EB>`, MB hashes become `HashOf<MB>`. CAS payload hashes (`set_transactions` return, `transactions(payload_hash)`) stay raw H256 since they're content-addressed, not entity-addressed. The cascade is real — ~30 files touched across ethexe/common, ethexe/db, ethexe/compute, ethexe/consensus, ethexe/malachite/service, ethexe/network, ethexe/observer, ethexe/prometheus, ethexe/rpc and ethexe/service. Where a function receives raw H256 from an external source (chain events, malachite-core API, test fixtures) we still cross the typed boundary with a local `mb_h` / `eb_h` helper rather than making `HashOf::new` safe, so each conversion remains explicit and auditable. `Mirror.processOutgoingAction` flow and existing tests continue to compile and lint clean across the workspace.
…hash_of/typed_hash
Eliminate the inner()/eb_h() ping-pong introduced when the typed-hash
refactor stopped at trait signatures: now the struct field itself
carries `HashOf<EB>`, so producers, consumers, and DB callers stay
typed end-to-end and only convert at the legitimate boundaries
(alloy event decode, malachite-core wire payload, ABI encoding).
- SimpleBlockData.hash / BlockData.hash → HashOf<EB>
- BlockFullData.hash (mock) → HashOf<EB>
- ChainCommitment.last_advanced_eth_block → HashOf<EB>
- BatchCommitment.block_hash → HashOf<EB>
- CommitmentSubmitted.block_hash → HashOf<EB>
- Transaction::AdvanceTillEthereumBlock.block_hash → HashOf<EB>
- PendingEvent.prerequisite → HashOf<EB>
- ConsensusService::receive_synced_block / receive_prepared_block → HashOf<EB>
- ObserverEvent::BlockSynced → HashOf<EB>
- BlockLoader::load / load_many input/output → HashOf<EB>
- ComputeEvent::BlockPrepared / Event::BlockPrepared → HashOf<EB>
- quarantine::{anchor, verify_passed, is_strict_descendant_of} → HashOf<EB>
- TxValidityChecker.start_block_hash → HashOf<EB>
- consensus::utils::is_eth_block_canonical_to → HashOf<EB>
- Compute error variants (BlockNotSynced, AdvanceMissingHeader, …) → HashOf<EB>
Removed: eb_hash_of(), SimpleBlockData::typed_hash(), and several
per-file `fn eb_h(h: H256) -> HashOf<EB>` helpers that became dead
code now that the source data is already typed. Helpers stay only
where they wrap genuinely untyped input (alloy event decode in
observer, user-supplied reference_block from injected txs, etc.).
EXPECTED_TYPE_INFO_HASH in db.rs updated to match the new layout.
All 12 ethexe-* crates compile (lib + tests). 420/420 nextest-run
tests pass (ethexe-common 16, malachite-core 192, observer/rpc/cli/
network/consensus/prometheus 212).
grishasobol
force-pushed
the
gsobol/ethexe/malachite-typed-block-hashes
branch
from
16ced2a to
8411f43
Compare
…erator Node fields with HashOf<EB>/<MB>
Cascade typed-hash coverage:
- `BlockHeader.parent_hash`: H256 → HashOf<EB>
- `InjectedTransaction.reference_block`: H256 → HashOf<EB>
- `iterator.rs` Node fields:
- ChainNode.head / .bottom → HashOf<EB>
- BlockNode / BlockMetaNode / BlockHeaderNode / BlockEventsNode /
BlockSyncedNode `.block` → HashOf<EB>
- MbNode / MbMetaNode / MbProgramStatesNode / MbScheduleNode /
MbScheduleTasksNode / MbOutcomeNode `.mb_hash` → HashOf<MB>
- `DatabaseIteratorError::NoBlock*` → HashOf<EB>, `::NoMb*` → HashOf<MB>
- `IntegrityVerifier` (visitor + verify_chain + fields): same retype
- Visitor signatures (visit_block_meta / visit_block_synced /
visit_block_header / visit_mb_schedule_tasks): typed args
Also:
- Drop the iterator `with_hash_eb`/`with_hash_mb` macro branches (the
storage trait already takes typed hashes, so `with_hash:` covers both).
- Remove the `unsafe { HashOf::<EB>::new(header.parent_hash) }`
ping-pongs across mempool/quarantine/tx_validity/observer/compute/
consensus/cli/service.
- Make `HashOf::<T>::zero()` const so it usable in `const` contexts
(`ProtocolTimelines` test fixture).
- Derive `MaxEncodedLen` on `HashOf<T>`, required by
`InjectedTransaction#[derive(MaxEncodedLen)]`.
EXPECTED_TYPE_INFO_HASH bumped: 04e137b4… → 26ef9eb1…
The `mb_h(H256) -> HashOf<MB>` helpers stay because malachite-core's
`Block::hash()` returns raw H256 — that's the wire boundary. Per-file
unused `eb_h` helpers in mempool/tx_validity are gated behind
`#[cfg(test)]` since prod paths now consume HashOf<EB> directly.
Tests: 383/383 pass across ethexe-common, ethexe-db, ethexe-malachite,
ethexe-malachite-core, ethexe-network, ethexe-rpc, ethexe-consensus,
ethexe-compute, ethexe-cli, ethexe-observer, ethexe-processor.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
ethexe_malachite_core::Block→ethexe_common::malachite::MB), the versionedBlockPayload, andCompactMbintoethexe-commonso the executor crates can use them without depending on the consensus layer.CompactMbnow matchesMB's shape with the opaque payload replaced bypayload_hash(renamed fromtransactions_hash).HashOf<MB>onCompactMb.parent,BlockMeta.last_committed_mb,DBGlobals.{latest_finalized_mb_hash, latest_computed_mb_hash};HashOf<EB>onMbMeta.last_advanced_eb,BlockMeta.last_committed_eb,DBGlobals.{start_block_hash, latest_prepared_eb_hash},DBConfig.genesis_block_hash.pub type EB = SimpleBlockData+SimpleBlockData::typed_hash()/eb_hash_of(H256)to lift the bare Ethereum block hash into the typed wrapper.HashOf<T>isEncode + Decodeas a bare 32-byte hash (phantom is#[codec(skip)]), so the on-wire/on-disk layout is byte-identical to before — only the static type at the field changes.MbStorageRO/RWtrait method parameters intentionally keepmb_hash: H256. ThreadingHashOf<MB>through every call site in compute/cli/consensus/processor/etc. would cascade through dozens of files for no extra safety gain — the boundary structs already catch the EB-vs-MB confusion that motivated the refactor. Call sites passfield.inner()where needed.Fixes #5507.
Test plan
cargo nextest run -p ethexe-common -p ethexe-db -p ethexe-malachite-core -p ethexe-malachite -p ethexe-compute -p ethexe-consensus— 248/248 pass (incl. multi-validator integration tests).cargo fmt,cargo clippy --workspace --tests— clean.ensure_types_unchangedupdated with the newEXPECTED_TYPE_INFO_HASH(CompactMb shape changed).🤖 Generated with Claude Code