feat(client): l1<>l2 messaging crate

[azurwastaken]

Pull Request type

• Feature

What is the current behavior?

Resolves: #204

What is the new behavior?

• Added a crate with a worker that listen for LogMessageToL2 event on main contract and handle it as the old madara did.
• Added some entries to the db for messaging related data

Does this introduce a breaking change?

No

Usefull information

🚨Require #208 🚨
This PR include a lot of code from #208 as we started from it to build with the feature from the eth crate.

A few notes on the current design:

• There is no guarantee that L1->L2 messages will be executed in a centralized context.
• There is no ordering guarantee, nonces are solely used

Current Progress

Overall Workflow

• Create a stream of LogMessageToL2 events
• Get last synced block from Messaging DB
• Consume the stream and log event
• Process message
  • Parse tx fee
  • Parse transaction from event
  • Check if message has already been processed
  • Build transaction
  • [Waiting for mempool Migrate to blockifier v0.8 and Sequencing / mempool part 1 #198 ] Submit tx to mempool
  • Update Messaging DB

Specific case to handle

• Handle Message Cancellation (feat: handle message cancellations on L1 keep-starknet-strange/madara#1638)

Tests

• E2E test 1

  • Launch Anvil Node
  • Launch Worker
  • Send L1->L2 message
  • Assert that event is emitted on L1
  • Assert that even was caught by the worker with correct data
  • Assert the tx hash computed by the worker is correct
  • Assert that the tx has been included in the mempool
  • Assert that DB was correctly updated (last synced block & nonce)
  • Assert that the tx was correctly executed

• E2E test 2

  • Should fail if we try to send multiple messages with same nonces

• E2E test 3

  • Message Cancellation

[Mohiiit]

Hey @azurwastaken, we were thinking of a different approach about the structure of the code. So instead of creating l1-messaging a different crate, we can have it inside the eth-crate, just like update_state. And as of now we are joining l1_sync with l1_messages_sync and l2_sync, so I was thinking to join l1_sync and l2_sync and inside the l1_sync we will have l1_messages_sync, state_update_sync and l1_gas_price sync, wdyt?

so structure of the code we had in mind was:

• eth-crate
  |_____ state_update
  |_____ l1_messages
  |_____ l1_gas_price
  |_____ maybe a sync.rs with sync function which calls above 3

and the above sync function get called in the sync crate, inside the starknet_sync_worker, wdyt?

cc: @cchudant

[azurwastaken]

Hey @azurwastaken, we were thinking of a different approach about the structure of the code. So instead of creating l1-messaging a different crate, we can have it inside the eth-crate, just like update_state. And as of now we are joining l1_sync with l1_messages_sync and l2_sync, so I was thinking to join l1_sync and l2_sync and inside the l1_sync we will have l1_messages_sync, state_update_sync and l1_gas_price sync, wdyt?

so structure of the code we had in mind was:

• eth-crate
  |_____ state_update
  |_____ l1_messages
  |_____ l1_gas_price
  |_____ maybe a sync.rs with sync function which calls above 3

and the above sync function get called in the sync crate, inside the starknet_sync_worker, wdyt?

cc: @cchudant

Hey @Mohiiit , yes you are right, its porbably better if everything is together

[cchudant]

Hey @azurwastaken, we were thinking of a different approach about the structure of the code. So instead of creating l1-messaging a different crate, we can have it inside the eth-crate, just like update_state. And as of now we are joining l1_sync with l1_messages_sync and l2_sync, so I was thinking to join l1_sync and l2_sync and inside the l1_sync we will have l1_messages_sync, state_update_sync and l1_gas_price sync, wdyt?

so structure of the code we had in mind was:

* eth-crate
  |_____ state_update
  |_____ l1_messages
  |_____ l1_gas_price
  |_____ maybe a `sync.rs` with sync function which calls above 3

and the above sync function get called in the sync crate, inside the starknet_sync_worker, wdyt?

cc: @cchudant

yes that makes sense
however a small thing here: i dont want the eth crate to be called from sync, i'd like it to be a Service so that i can just activate it even when doing block production (not sync) mode

also in the future this eth crate will expose a common interface to the rest of the app that is swappable with other chains (e.g for starknet L3s) but i'm getting ahead of myself

[azurwastaken]

Outdated, see #220