Create a ChainSpec Wrapper to manage Anvil genesis config #293

re-gius · 2025-09-09T17:15:57Z

Support genesis configuration customization for anvil-polkadot.
This is implemented through a wrapper struct around ChainSpec, called DevelopmentChainSpec, which re-implements assimilate_storage for the BuildStorage trait to override genesis storage for customizable items: chain id, block number, and timestamp.

For reference, also see #260 (comment)

Notice that there are still some missing genesis storage items to override: alloc (accounts with code/balances), coinbase (block author), gas settings.

crates/anvil-polkadot/src/substrate_node/genesis.rs

re-gius · 2025-09-17T15:20:55Z

EDIT: actually, I don't see it's producing blocks, what I was seeing was probably the anvil config printed as string, not the genesis block storage. So I would say it's not working yet

alindima · 2025-09-18T10:22:34Z

EDIT: actually, I don't see it's producing blocks, what I was seeing was probably the anvil config printed as string, not the genesis block storage. So I would say it's not working yet

After Alex's block production PR, we no longer start producing blocks by default. You either need to configure it via RPC or CLI. A simple way is to use the -b <block_interval> CLI param, which will produce a block every block_interval seconds

crates/anvil-polkadot/src/genesis.rs

alindima · 2025-09-18T10:32:22Z

crates/anvil-polkadot/src/substrate_node/genesis.rs

+}
+
+#[cfg(test)]
+mod tests {


these tests are not really useful

Changed in 2ff0438 to test for key-value storage translation, can they be useful now?

I think we should be using the hardcoded variants as defined in this module. After all, this is what we use in reality

let's also do this

Fixed in e1abde2

crates/anvil-polkadot/src/opts.rs

crates/anvil-polkadot/src/substrate_node/chain_spec.rs

crates/anvil-polkadot/Cargo.toml

crates/anvil-polkadot/src/genesis.rs

alindima · 2025-09-18T15:16:41Z

I tried customising the genesis block number but without success. When querying the numer from polkadot.js, I get <unknown>. Could it be because you are encoding it as u64 instead of u32?

Anyway, setting it in the genesis state is not going to be enough. The block builder uses the block number from the header of the genesis block (not from the runtime state). I think we need to implement another GenesisBlockBuilder (by implementing BuildGenesisBlock on a newtype), which uses the configured block number. See the construct_genesis_block function, this is the place where we set the block number to 0.

alindima · 2025-09-18T15:36:45Z

And for timestamp, we also need to modify the timestamp of the mining engine (similar to what the setTime RPC is doing), so that the next blocks will be based on this timestamp (otherwise they'll just jump right back to the present moment on the next block). CC: @AlexandruCihodaru

iulianbarbu · 2025-09-19T08:29:36Z

And for timestamp, we also need to modify the timestamp of the mining engine (similar to what the setTime RPC is doing), so that the next blocks will be based on this timestamp (otherwise they'll just jump right back to the present moment on the next block). CC: @AlexandruCihodaru

Yeh, looks like block 1 references current time. I don't think we're able to tell what timestamp does block 0 reference, with polkadotjs (no extrinsic shows up in it, like in block 1 case, and the header doesn't seem to show this info either). Would be nice. Not sure how subscan shows it for the production networks.

iulianbarbu · 2025-09-19T08:31:26Z

Anyway, setting it in the genesis state is not going to be enough. The block builder uses the block number from the header of the genesis block (not from the runtime state).

I am thinking we should set the genesis block number in the genesis block header too. Not sure though how doable it is.
LE: maybe that means we'll not need to reimplement the genesis block builder. On the plus side the info shown in polkadotjs will be in accordance with how the node was started.

crates/anvil-polkadot/tests/it/genesis.rs

iulianbarbu

LGTM, nice work! Would be great to have also the chain id tested like Alin mentioned offline (which IIUC should be easily doable).

crates/anvil-polkadot/tests/it/genesis.rs

Co-authored-by: Alin Dima <[email protected]>

AlexandruCihodaru · 2025-09-29T13:51:21Z

crates/anvil-polkadot/src/substrate_node/service.rs

-    let time_manager =
-        Arc::new(TimeManager::new_with_milliseconds(sp_timestamp::Timestamp::current().into()));
+    let time_manager = Arc::new(TimeManager::new_with_milliseconds(
+        sp_timestamp::Timestamp::from(anvil_config.get_genesis_timestamp()).into(),


get_genesis_timestamp will return the current time as seconds, but we create the TimeManager based on milliseconds.

I had to use --block-time 3 for 3 seconds block interval. The CLI and the behavior during testing corresponds. Somewhere a seconds to milliseconds conversion must be done I think.

It is done, all parameters to RPCs are converted to milliseconds by the TimeManager methods.

Thanks! Changed in 90da646

& fixed accordingly in 7221e82

alindima

Thanks! Nice work!

Approving, modulo Alex's comment and let's make test_genesis less flaky in the CI

AlexandruCihodaru · 2025-09-29T16:32:08Z

crates/anvil-polkadot/src/substrate_node/genesis.rs

+impl<'a> From<&'a AnvilNodeConfig> for GenesisConfig {
+    fn from(anvil_config: &'a AnvilNodeConfig) -> Self {
+        Self {
+            chain_id: anvil_config.get_chain_id(),


I think we should also update standard_rpc::test_get_chain_id

yes, we should properly implement the get_chain_id RPC to read the state instead of returning the hardcoded constant

AlexandruCihodaru · 2025-09-29T16:33:09Z

crates/anvil-polkadot/tests/it/genesis.rs

+    let genesis_hash = node.block_hash_by_number(genesis_block_number).await.unwrap();
+    let actual_genesis_timestamp = node.get_decoded_timestamp(Some(genesis_hash)).await;
+    // Anvil genesis timestamp is in seconds, while Substrate timestamp is in milliseconds.
+    assert_eq!(actual_genesis_timestamp, genesis_timestamp * 1000);


nit: use checked_mul

Changed in fcf200b

AlexandruCihodaru · 2025-09-29T17:28:26Z

crates/anvil-polkadot/tests/it/genesis.rs

+
+#[tokio::test(flavor = "multi_thread")]
+async fn test_genesis() {
+    let genesis_block_number: u32 = 1000;


When creating the OnlineClient in create_revive_rpc_client

foundry-polkadot/crates/anvil-polkadot/src/api_server/server.rs

Line 414 in a65a8e1

let api = OnlineClient::<SrcChainConfig>::from_rpc_client(rpc_client.clone()).await?;

there is this assumption that genesis block number is 0
https://github.com/paritytech/subxt/blob/884dfe6c69753584660f53352da636f9d0706631/rpcs/src/methods/legacy.rs#L89-L94

is there a problem with this? Did you run into any issues?
If not, then that's fine. We shouldn't need to call this for our use case

if we set the genesis number to 1000 as in the test, we will not be able to create the OnlineClient because it calls the get_genesis from subxt that I linked above, which assumes that the genesis number is zero

ah I see. I think there's an easy workaround. Instead of calling OnlineClient::<SrcChainConfig>::from_rpc_client we should call from_rpc_client_with() and pass the genesis hash as returned by our local backend

Changed in fcf200b

AlexandruCihodaru · 2025-10-01T14:40:45Z

crates/anvil-polkadot/src/substrate_node/genesis.rs

+            // Anvil genesis timestamp is in seconds, while Substrate timestamp is in milliseconds.
+            timestamp: anvil_config.get_genesis_timestamp() * 1000,
+            alloc: anvil_config.genesis.as_ref().map(|g| g.alloc.clone()),
+            number: anvil_config.get_genesis_number() as u32,


This could fail 🤷

Added expect in e1abde2

AlexandruCihodaru · 2025-10-01T14:41:04Z

crates/anvil-polkadot/src/substrate_node/genesis.rs

+        Self {
+            chain_id: anvil_config.get_chain_id(),
+            // Anvil genesis timestamp is in seconds, while Substrate timestamp is in milliseconds.
+            timestamp: anvil_config.get_genesis_timestamp() * 1000,


Maybe use checked_mul here?

Changed in e1abde2

AlexandruCihodaru · 2025-10-01T14:41:26Z

crates/anvil-polkadot/src/api_server/server.rs

+        .client
+        .storage(genesis_hash, &StorageKey(CODE_KEY.to_vec()))
+        .expect("Runtime code not found for given genesis hash")
+        .unwrap();


Please do proper error handling

AlexandruCihodaru · 2025-10-01T14:42:28Z

crates/anvil-polkadot/src/api_server/server.rs

+    let runtime_version = substrate_service
+        .client
+        .runtime_version_at(genesis_hash)
+        .expect("Runtime version not found for given genesis hash");


Let's propagate the error instead of expect

AlexandruCihodaru · 2025-10-01T15:25:33Z

crates/anvil-polkadot/tests/it/genesis.rs

+    assert_eq!(current_chain_id, chain_id);
+
+    // Manually mine two blocks and force the timestamp to be increasing with 1 second each time.
+    unwrap_response::<()>(


The timestamp of the second block will include everything that happens from when the TimeManager is created until the inherent is created, so maybe take a larger tolerance in the assert

AlexandruCihodaru · 2025-10-01T15:25:51Z

crates/anvil-polkadot/src/substrate_node/service.rs

-    let time_manager =
-        Arc::new(TimeManager::new_with_milliseconds(sp_timestamp::Timestamp::current().into()));
+    let time_manager = Arc::new(TimeManager::new_with_milliseconds(
+        sp_timestamp::Timestamp::from(1000 * anvil_config.get_genesis_timestamp()).into(),


nit: checked_mul

Changed in e1abde2

AlexandruCihodaru · 2025-10-01T15:27:42Z

crates/anvil-polkadot/src/api_server/server.rs

+    let api = OnlineClient::<SrcChainConfig>::from_rpc_client_with(
+        genesis_hash,
+        subxt_runtime_version,
+        subxt_metadata,
+        rpc_client.clone(),
+    )?;


nit: move the creation of the OnlineClient in a separate method

I think it belongs here, we're after all only creating this for the purposes of the revive rpc client

alindima · 2025-10-06T14:18:39Z

crates/anvil-polkadot/src/api_server/server.rs

+    // Using best_hash because genesis hash is only set if genesis block number is equal to 0, but
+    // Anvil allows a custom genesis block number.
+    // https://github.com/paritytech/polkadot-sdk/blob/62c3a80f913a272c4f5dba2c91320056d39ec68e/substrate/client/api/src/in_mem.rs#L178
+    let genesis_hash = substrate_service.client.chain_info().best_hash;


This will probably work in most cases but it's a bit racy.

The genesis hash here may be different if the node is configured to mine blocks very frequently, it could get to mine a block before we retrieve the first block's hash here.

The clean solution would be to retrieve the block hash of the genesis block number (which we know).
Let's do this instead

Actually, the genesis_hash set in from_rpc_client_with will probably need to be equal to the zeroed hash (which is what the client returns).
AFAIK the extrinsic payload contains the genesis hash as well (being constructed by subxt), so the runtime will reject the transaction if the hash is different.

Let's add a transaction to the integration test and see what happens and work from there

This is why I see tests failing in CI with:

called `Result::unwrap()` on an `Err` value: "Expected success but got error: RpcError { code: InternalError, message: \"Client error: RPC error: RPC error: client error: ErrorObject { code: ServerError(1010), message: \\\"Invalid Transaction\\\", data: Some(RawValue(\\\"Transaction call is not expected\\\")) }\", data: None }"

alindima · 2025-10-06T14:25:18Z

crates/anvil-polkadot/tests/it/genesis.rs

+use anvil_polkadot::config::{AnvilNodeConfig, SubstrateNodeConfig};
+
+#[tokio::test(flavor = "multi_thread")]
+async fn test_genesis() {


we should also add a test for configuring via json file, but this can come later, in the next PR if you want

re-gius added 3 commits September 9, 2025 19:14

draft chainspec wrapper

6bc93a2

inject Anvil Node Config into Substrate CLI

38abb13

Set chain_id from config.json

9b33fb1

iulianbarbu reviewed Sep 15, 2025

View reviewed changes

crates/anvil-polkadot/src/substrate_node/genesis.rs Show resolved Hide resolved

iulianbarbu linked an issue Sep 15, 2025 that may be closed by this pull request

[Anvil] Genesis customisation #241

Open

iulianbarbu mentioned this pull request Sep 15, 2025

[Anvil] State injector #240

Open

re-gius added 2 commits September 15, 2025 18:20

Add extra fields

30717d7

Merge branch 'master' into re-gius/chainspec-wrapper

66ee492

iulianbarbu mentioned this pull request Sep 16, 2025

[Anvil] Genesis customisation #241

Open

re-gius added 3 commits September 16, 2025 18:28

inject storage items into ChainSpec

85527e4

implement ChainSpec trait for wrapper

b4420d0

Merge branch 'master' into re-gius/chainspec-wrapper

b592f9a

re-gius marked this pull request as ready for review September 17, 2025 15:17

re-gius requested review from alindima and iulianbarbu September 17, 2025 15:17

fix clippy

f807562

alindima reviewed Sep 18, 2025

View reviewed changes

crates/anvil-polkadot/Cargo.toml Outdated Show resolved Hide resolved

re-gius added 2 commits September 18, 2025 16:21

CR reorg

2ff0438

Merge branch 'master' into re-gius/chainspec-wrapper

2302d04

alindima reviewed Sep 18, 2025

View reviewed changes

crates/anvil-polkadot/src/genesis.rs Outdated Show resolved Hide resolved

re-gius added 3 commits September 19, 2025 16:35

Merge branch 'master' into re-gius/chainspec-wrapper

56b192f

set System::Number as u32

26e0fe7

nit

7caa8a5

re-gius added 2 commits September 26, 2025 16:49

CR nits + reorg

ce9de23

Add genesis integration tests (chain ID missing)

6e67186

alindima reviewed Sep 29, 2025

View reviewed changes

crates/anvil-polkadot/tests/it/genesis.rs Outdated Show resolved Hide resolved

iulianbarbu approved these changes Sep 29, 2025

View reviewed changes

crates/anvil-polkadot/tests/it/genesis.rs Show resolved Hide resolved

crates/anvil-polkadot/tests/it/genesis.rs Outdated Show resolved Hide resolved

re-gius and others added 3 commits September 29, 2025 12:28

comment nit

e2b2072

Co-authored-by: Alin Dima <[email protected]>

add chain id test

3ee3857

Merge branch 'master' into re-gius/chainspec-wrapper

3bd0395

re-gius self-assigned this Sep 29, 2025

AlexandruCihodaru reviewed Sep 29, 2025

View reviewed changes

alindima approved these changes Sep 29, 2025

View reviewed changes

re-gius added 5 commits September 29, 2025 16:04

fix time manager creation

90da646

changed genesis milliseconds mismatch + clippy for unused import

7221e82

undo clippy change

867d680

Merge branch 'master' into re-gius/chainspec-wrapper

327b973

merge fix - missing import

7680f00

AlexandruCihodaru reviewed Sep 29, 2025

View reviewed changes

re-gius added 3 commits October 1, 2025 15:35

fix rpc client creation

fcf200b

clippy nit

5e42322

fix

c51999b

AlexandruCihodaru reviewed Oct 1, 2025

View reviewed changes

Merge branch 'master' into re-gius/chainspec-wrapper

f8ba53b

alindima reviewed Oct 6, 2025

View reviewed changes

re-gius added 4 commits October 7, 2025 17:56

Merge branch 'master' into re-gius/chainspec-wrapper

560962e

CR fixes

e1abde2

fix types

c79e41b

nit

c36ee17

Create a ChainSpec Wrapper to manage Anvil genesis config #293

Are you sure you want to change the base?

Create a ChainSpec Wrapper to manage Anvil genesis config #293

Conversation

re-gius commented Sep 9, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

re-gius commented Sep 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

alindima commented Sep 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

alindima commented Sep 18, 2025

Uh oh!

alindima commented Sep 18, 2025

Uh oh!

iulianbarbu commented Sep 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

iulianbarbu commented Sep 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

iulianbarbu left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

re-gius Sep 29, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

alindima left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

AlexandruCihodaru Sep 29, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

re-gius commented Sep 9, 2025 •

edited

Loading

re-gius commented Sep 17, 2025 •

edited

Loading

alindima commented Sep 18, 2025 •

edited

Loading

iulianbarbu commented Sep 19, 2025 •

edited

Loading

iulianbarbu commented Sep 19, 2025 •

edited

Loading

re-gius Sep 29, 2025 •

edited

Loading

AlexandruCihodaru Sep 29, 2025 •

edited

Loading

alindima Oct 6, 2025 •

edited

Loading

alindima Oct 6, 2025 •

edited

Loading