Michal Kucharczyk
6a295e7c28
* substrate-test-runtime migrated to pure-frame based * test block builder: helpers added * simple renaming * basic_authorship test adjusted * block_building storage_proof test adjusted * babe: tests: should_panic expected added * babe: tests adjusted ConsensusLog::NextEpochData is now added by pallet_babe as pallet_babe::SameAuthoritiesForever trigger is used in runtime config. * beefy: tests adjusted test-substrate-runtime is now using frame::executive to finalize the block. during finalization the digests stored during block execution are checked against header digests: https://github.com/paritytech/substrate/blob/91bb2d29ca905599098a5b35eaf24867c4fbd60a/frame/executive/src/lib.rs#L585-L591 It makes impossible to directly manipulate header's digets, w/o depositing logs into system pallet storage `Digest<T: Config>`. Instead of this dedicated extrinsic allowing to store logs items (MmrRoot / AuthoritiesChange) is used. * grandpa: tests adjusted test-substrate-runtime is now using frame::executive to finalize the block. during finalization the digest logs stored during block execution are checked against header digest logs: https://github.com/paritytech/substrate/blob/91bb2d29ca905599098a5b35eaf24867c4fbd60a/frame/executive/src/lib.rs#L585-L591 It makes impossible to directly manipulate header's digets, w/o depositing logs into system pallet storage `Digest<T: Config>`. Instead of this dedicated extrinsic allowing to store logs items (ScheduledChange / ForcedChange and DigestItem::Other) is used. * network:bitswap: test adjusted The size of unchecked extrinsic was increased. The pattern used in test will be placed at the end of scale-encoded buffer. * runtime apis versions adjusted * storage keys used in runtime adjusted * wasm vs native tests removed * rpc tests: adjusted Transfer transaction processing was slightly improved, test was adjusted. * tests: sizes adjusted Runtime extrinsic size was increased. Size of data read during block execution was also increased due to usage of new pallets in runtime. Sizes were adjusted in tests. * cargo.lock update cargo update -p substrate-test-runtime -p substrate-test-runtime-client * warnings fixed * builders cleanup: includes / std * extrinsic validation cleanup * txpool: benches performance fixed * fmt * spelling * Apply suggestions from code review Co-authored-by: Davide Galassi <davxy@datawok.net> * Apply code review suggestions * Apply code review suggestions * get rid of 1063 const * renaming: UncheckedExtrinsic -> Extrinsic * test-utils-runtime: further step to pure-frame * basic-authorship: tests OK * CheckSubstrateCall added + tests fixes * test::Transfer call removed * priority / propagate / no sudo+root-testing * fixing warnings + format * cleanup: build2/nonce + format * final tests fixes all tests are passing * logs/comments removal * should_not_accept_old_signatures test removed * make txpool benches work again * Cargo.lock reset * format * sudo hack removed * txpool benches fix+cleanup * .gitignore reverted * rebase fixing + unsigned cleanup * Cargo.toml/Cargo.lock cleanup * force-debug feature removed * mmr tests fixed * make cargo-clippy happy * network sync test uses unsigned extrinsic * cleanup * ".git/.scripts/commands/fmt/fmt.sh" * push_storage_change signed call remove * GenesisConfig cleanup * fix * fix * GenesisConfig simplified * storage_keys_works: reworked * storage_keys_works: expected keys in vec * storage keys list moved to substrate-test-runtime * substrate-test: some sanity tests + GenesisConfigBuilder rework * Apply suggestions from code review Co-authored-by: Bastian Köcher <git@kchr.de> * Apply suggestions from code review * Review suggestions * fix * fix * beefy: generate_blocks_and_sync block_num sync with actaul value * Apply suggestions from code review Co-authored-by: Davide Galassi <davxy@datawok.net> * Update test-utils/runtime/src/genesismap.rs Co-authored-by: Davide Galassi <davxy@datawok.net> * cargo update -p sc-rpc -p sc-transaction-pool * Review suggestions * fix * doc added * slot_duration adjusted for Babe::slot_duration * small doc fixes * array_bytes::hex used instead of hex * tiny -> medium name fix * Apply suggestions from code review Co-authored-by: Sebastian Kunert <skunert49@gmail.com> * TransferData::try_from_unchecked_extrinsic -> try_from * Update Cargo.lock --------- Co-authored-by: parity-processbot <> Co-authored-by: Davide Galassi <davxy@datawok.net> Co-authored-by: Bastian Köcher <git@kchr.de> Co-authored-by: Sebastian Kunert <skunert49@gmail.com>
BABE (Blind Assignment for Blockchain Extension)
BABE is a slot-based block production mechanism which uses a VRF PRNG to randomly perform the slot allocation. On every slot, all the authorities generate a new random number with the VRF function and if it is lower than a given threshold (which is proportional to their weight/stake) they have a right to produce a block. The proof of the VRF function execution will be used by other peer to validate the legitimacy of the slot claim.
The engine is also responsible for collecting entropy on-chain which will be used to seed the given VRF PRNG. An epoch is a contiguous number of slots under which we will be using the same authority set. During an epoch all VRF outputs produced as a result of block production will be collected on an on-chain randomness pool. Epoch changes are announced one epoch in advance, i.e. when ending epoch N, we announce the parameters (randomness, authorities, etc.) for epoch N+2.
Since the slot assignment is randomized, it is possible that a slot is assigned to multiple validators in which case we will have a temporary fork, or that a slot is assigned to no validator in which case no block is produced. Which means that block times are not deterministic.
The protocol has a parameter c [0, 1] for which 1 - c is the probability
of a slot being empty. The choice of this parameter affects the security of
the protocol relating to maximum tolerable network delays.
In addition to the VRF-based slot assignment described above, which we will call primary slots, the engine also supports a deterministic secondary slot assignment. Primary slots take precedence over secondary slots, when authoring the node starts by trying to claim a primary slot and falls back to a secondary slot claim attempt. The secondary slot assignment is done by picking the authority at index:
blake2_256(epoch_randomness ++ slot_number) % authorities_len.
The secondary slots supports either a SecondaryPlain or SecondaryVRF
variant. Comparing with SecondaryPlain variant, the SecondaryVRF variant
generates an additional VRF output. The output is not included in beacon
randomness, but can be consumed by parachains.
The fork choice rule is weight-based, where weight equals the number of primary blocks in the chain. We will pick the heaviest chain (more primary blocks) and will go with the longest one in case of a tie.
An in-depth description and analysis of the protocol can be found here: https://research.web3.foundation/en/latest/polkadot/block-production/Babe.html
License: GPL-3.0-or-later WITH Classpath-exception-2.0