fix(anvil): reset from fork to another fork

[yash-atreya]

Motivation

Fixes #8684

Resetting from an existing fork to another fork would not update underlying ForkedDatabase, resulting in an invalid state for the new fork.

Solution

reset the underlying the ForkedDatabase using node_config.setup_fork_db_and_config

[mattsse]

why do we no longer need this?

[yash-atreya]

Correct me if I'm wrong but we no longer need this because when we reset to a new forked state from a non-fork/old-fork state, we use self.db(Backend.db) as an intermediate step to set the database in ClientFork, here and here.

So even when we clear - self.db.write().await.clear();, RPC calls work because they refer to the database in the fork + foundry_fork_db::SharedBackend.

[grandizzy]

hey @yash-atreya, wonder if you got the chance to test it with arbitrum forks too, there's a special case with block L1/L2 Not sure if it needs to be addressed here too but think it worth flagging

foundry/crates/anvil/src/eth/backend/mem/mod.rs

Lines 1734 to 1746 in d127137

	// If Arbitrum, apply chain specifics to converted block.
	if let Ok(
	NamedChain::Arbitrum |
	NamedChain::ArbitrumGoerli |
	NamedChain::ArbitrumNova |
	NamedChain::ArbitrumTestnet,
	) = NamedChain::try_from(self.env.read().env.cfg.chain_id)
	{
	// Block number is the best number.
	block.header.number = Some(self.best_number());
	// Set `l1BlockNumber` field.
	block.other.insert("l1BlockNumber".to_string(), number.into());
	}