# Bridge Parachains Pallet

The bridge parachains pallet is a light client for one or several parachains of the bridged relay chain.
It serves as a source of finalized parachain headers and is used when you need to build a bridge with
a parachain.

The pallet requires [bridge GRANDPA pallet](../grandpa/) to be deployed at the same chain - it is used
to verify storage proofs, generated at the bridged relay chain.

## A Brief Introduction into Parachains Finality

You can find detailed information on parachains finality in the [Polkadot](https://github.com/paritytech/polkadot)
and [Cumulus](https://github.com/paritytech/cumulus) repositories. This section gives a brief overview of how
the parachain finality works and how to build a light client for a parachain.

The main thing there is that the parachain generates blocks on its own, but it can't achieve finality without
help of its relay chain. Instead, the parachain collators create a block and hand it over to the relay chain
validators. Validators validate the block and register the new parachain head in the
[`Heads` map](https://github.com/paritytech/polkadot/blob/88013730166ba90745ae7c9eb3e0c1be1513c7cc/runtime/parachains/src/paras/mod.rs#L645)
of the [`paras`](https://github.com/paritytech/polkadot/tree/master/runtime/parachains/src/paras) pallet,
deployed at the relay chain. Keep in mind that this pallet, deployed at a relay chain, is **NOT** a bridge pallet,
even though the names are similar. 

And what the bridge parachains pallet does, is simply verifying storage proofs of parachain heads within that
`Heads` map. It does that using relay chain header, that has been previously imported by the
[bridge GRANDPA pallet](../grandpa/). Once the proof is verified, the pallet knows that the given parachain
header has been finalized by the relay chain. The parachain header fields may then be used to verify storage
proofs, coming from the parachain. This allows the pallet to be used e.g. as a source of finality for the messages
pallet.

## Pallet Operations

The main entrypoint of the pallet is the `submit_parachain_heads` call. It has three arguments:

- storage proof of parachain heads from the `Heads` map;

- parachain identifiers and hashes of their heads from the storage proof;

- the relay block, at which the storage proof has been generated.

The pallet may track multiple parachains. And the parachains may use different primitives - one may use 128-bit block
numbers, other - 32-bit. To avoid extra decode operations, the pallet is using relay chain block number to order
parachain headers. Any finalized descendant of finalized relay block `RB`, which has parachain block `PB` in
its `Heads` map, is guaranteed to have either `PB`, or its descendant. So parachain block number grows with relay
block number.

The pallet may reject parachain head if it already knows better (or the same) head. In addition, pallet rejects
heads of untracked parachains.

The pallet doesn't track anything behind parachain heads. So it requires no initialization - it is ready to accept
headers right after deployment.

## Non-Essential Functionality

There may be a special account in every runtime where the bridge parachains module is deployed. This
account, named 'module owner', is like a module-level sudo account - he's able to halt and
resume all module operations without requiring runtime upgrade. Calls that are related to this
account are:

- `fn set_owner()`: current module owner may call it to transfer "ownership" to another account;

- `fn set_operating_mode()`: the module owner (or sudo account) may call this function to stop all
  module operations. After this call, all finality proofs will be rejected until further `set_operating_mode` call'.
  This call may be used when something extraordinary happens with the bridge.

If pallet owner is not defined, the governance may be used to make those calls.

## Signed Extension to Reject Obsolete Headers

It'd be better for anyone (for chain and for submitters) to reject all transactions that are submitting
already known parachain heads to the pallet. This way, we leave block space to other useful transactions and
we don't charge concurrent submitters for their honest actions.

To deal with that, we have a [signed extension](./src/extension.rs) that may be added to the runtime.
It does exactly what is required - rejects all transactions with already known heads. The submitter
pays nothing for such transactions - they're simply removed from the transaction pool, when the block
is built.

The signed extension, however, is a bit limited - it only works with transactions that provide single
parachain head. So it won't work with multiple parachain heads transactions. This fits our needs
for [Kusama <> Polkadot bridge](../../docs/polkadot-kusama-bridge-overview.md). If you need to deal
with other transaction formats, you may implement similar extension for your runtime.

You may also take a look at the [`generate_bridge_reject_obsolete_headers_and_messages`](../../bin/runtime-common/src/lib.rs)
macro that bundles several similar signed extensions in a single one.

## Parachains Finality Relay

We have an offchain actor, who is watching for new parachain heads and submits them to the bridged chain.
It is the parachains relay - you may look at the [crate level documentation and the code](../../relays/parachains/).