feat: Cosmos SDK Module #5370

yjamin · 2025-02-04T16:54:01Z

Description

This PR adds Agent support for the upcoming Hyperlane Cosmos SDK Module, called CosmosNative, to differentiate it from the existing CosmWasm (CW) implementation.

Drive-by Fixes

None

Related Issues

None reported.

Backward Compatibility

No breaking changes—this PR extends functionality without affecting existing implementations.

Testing

The CosmosNative implementation has been tested through:

Automated E2E tests – Following the same approach as the CW implementation.
Deployed preview – Currently live on the alpha network for additional validation.

E2E Testing Details

They pretty much follow the same structure as the CW tests.
They spin up a lightweight Cosmos SDK chain (simapp) with the Hyperlane Cosmos module.

Note: The simapp includes a preview version of the Hyperlane module since it's not officially released yet.

To run the tests locally, use:

cargo test -r -p run-locally --features cosmosnative -- cosmosnative::test --nocapture

changeset-bot · 2025-02-04T16:54:05Z

⚠️ No Changeset found

Latest commit: daf2b1e

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

tkporter

Only spent about 20 mins the other day poking around some of the first files, hopefully will have time to leave more feedback but posting for now

The biggest thing that stands out to me is just how much duplication there is with hyperlane-cosmos. I could see there being some arguments for keeping this separate but I feel like consolidating as much as possible is probably the better move. Curious what @daniel-savu @ameten think

tkporter · 2025-02-09T17:37:22Z

rust/main/chains/hyperlane-cosmos-native/src/indexers/dispatch.rs

+        let tip = self.get_finalized_block_number().await?;
+        let sequence = self
+            .provider
+            .rest()


When is rest vs grpc vs rpc appropriate? Given we have rpc and grpc for the CW implementation, I have a slight preference to just keep needing just those 2 - is that possible?

Rationale being that we have some work in progress rn to add some metrics for our Cosmos gRPC & RPC interactions, would be nice to not need to extend that work to cover rest too

In Cosmos, gRPC, REST, and RPC all communicate with the chain through ABCI. They serve as wrappers around raw ABCI queries, meaning the main differences is in convenience and probably some performance overhead.

We chose to remove gRPC since it is not fully supported on our public KYVE endpoints, but this shouldn't matter for the agents implementation. We used REST as the query client implementation, because of its simplicity and human readability.

If we want to stick with gRPC and RPC, we would need to manually implement the gRPC Query Client for the Cosmos module. This includes handling protobuf serialization for each struct, i think it would make sense to move this into a separate repository (similar to the CW queries) and add that to the monorepo.

tkporter · 2025-02-09T17:38:04Z

rust/main/chains/hyperlane-cosmos-native/src/indexers/dispatch.rs

+            let value = attribute
+                .value_str()
+                .map_err(HyperlaneCosmosError::from)?
+                .replace("\"", "");


why's this necessary? curious what a raw EventAttribute here would look like

tkporter · 2025-02-09T17:52:23Z

rust/main/chains/hyperlane-cosmos-native/src/ism.rs

+
+    async fn get_ism(&self) -> ChainResult<Option<ISM>> {
+        let isms = self.provider.rest().isms(ReorgPeriod::None).await?;
+        for ism in isms {


should be able to do .into_iter() to not need to clone

tkporter · 2025-02-09T17:52:45Z

rust/main/chains/hyperlane-cosmos-native/src/ism.rs

+    /// metadata offchain fetching and onchain formatting standard.
+    async fn module_type(&self) -> ChainResult<ModuleType> {
+        let isms = self.provider.rest().isms(ReorgPeriod::None).await?;
+        for ism in isms {


tkporter · 2025-02-09T22:20:35Z

rust/main/chains/hyperlane-cosmos-native/src/error.rs

+/// This error can then be converted into the broader error type
+/// in hyperlane-core using the `From` trait impl
+#[derive(Debug, thiserror::Error)]
+pub enum HyperlaneCosmosError {


thoughts on reusing this with hyperlane-cosmos?

tkporter · 2025-02-09T22:21:06Z

rust/main/chains/hyperlane-cosmos-native/src/indexers/indexer.rs

+
+#[derive(Debug, Eq, PartialEq)]
+/// An event parsed from the RPC response.
+pub struct ParsedEvent<T: PartialEq> {


lots of opportunity for reuse with hyperlane-cosmos

ameten · 2025-02-26T11:30:31Z

rust/main/chains/hyperlane-cosmos-native/Cargo.toml

+tracing = { workspace = true }
+tracing-futures = { workspace = true }
+url = { workspace = true }
+prost = "0.13.4"


I wonder if we can reference it in workspace?

Good point, will look into it

ameten · 2025-02-26T11:30:46Z

rust/main/chains/hyperlane-cosmos-native/Cargo.toml

+hex = { workspace = true }
+http = { workspace = true }
+hyperlane-core = { path = "../../hyperlane-core", features = ["async"] }
+hyperlane-cosmwasm-interface.workspace = true


Suggested change

hyperlane-cosmwasm-interface.workspace = true

hyperlane-cosmwasm-interface = { workspace = true }

ameten · 2025-02-26T11:31:25Z

rust/main/Cargo.toml

I wonder why some dependencies are remove. We don't use them anymore?

Not too sure. I only used the cli for managing dependencies. Can re-add them if we want to be sure.

ameten · 2025-02-26T11:33:37Z

rust/main/hyperlane-base/src/contract_sync/cursors/sequence_aware/forward.rs

@@ -110,7 +110,7 @@ impl<T: Debug + Clone + Sync + Send + Indexable + 'static> ForwardSequenceAwareS
    /// If there are no logs to index, returns `None`.
    /// If there are logs to index, returns the range of logs, either by sequence or block number
    /// depending on the mode.
-    #[instrument(ret)]
+    // #[instrument(ret)]


I wonder why this annotation was removed?

Sorry, just forgot to remove the comment again.

ameten · 2025-02-26T11:41:33Z

rust/main/hyperlane-base/src/contract_sync/cursors/sequence_aware/backward.rs

+        // limit this to 1
+        if start < 1 {
+            return 1..=current_indexing_snapshot.at_block;
+        }


I wonder if it will work for all chains. Presumably, Ethereum genesis block has height 0.

Yes, good point, that was just annoying when testing with a local chain.

This will probably never be the case in a real world deployment, however i think it would make sense to add an optional additional property like genesis_height or deployment_height in IndexSettings that defines on what height the mailbox has been deployed and this should be the lower bound of each indexer, as there can't be any events prior to that height.

The default value of that property can then be set by the protocol type.

ameten · 2025-02-26T11:47:55Z

rust/main/hyperlane-base/src/settings/chains.rs

Just for visibility: RoutingISM and AggregationISM are not implemented

Correct, they are not supported by the Cosmos Module implementation yet. Will add a comment for clarity 👍

ameten · 2025-02-26T12:04:14Z

rust/main/hyperlane-base/src/traits/checkpoint_syncer.rs

+        // always write the 0th index
+        if index > curr || index == 0 {


I wonder what is the issue which requires this change?

Can't seem to replicate the issue anymore, nor did i fully understand it at that time.
When i first started the implementation, the syncer wouldn't write the 0th index. That was probably due to a faulty implementation of latest_checkpoint in our merkle_tree_hook. A lot has changed since then, i would just remove it again.

ameten · 2025-02-26T12:09:34Z

rust/main/config/testnet_config.json

@@ -1515,7 +1515,7 @@
      "protocol": "ethereum",
      "rpcUrls": [
        {
-          "http": "http://127.0.0.1:8545"
+          "http": "http://0.0.0.0:8545"


I wonder if this change was intentional?

Nope, not intentional, will remove. Thanks 👍

ameten · 2025-02-26T12:21:51Z

rust/main/chains/hyperlane-cosmos-native/src/providers/cosmos.rs

I wonder if it will make sense to have the functionality shared by CosmosWasm and CosmosNative implementations to be extracted into separate module and re-used?

Example: sender_and_nonce method in CosmosProvider and CosmosNativeProvider. There can be more examples like this.

ameten · 2025-02-26T12:28:28Z

rust/main/chains/hyperlane-cosmos-native/src/providers/cosmos.rs

+use std::{io::Cursor, ops::Deref};
+
+use cosmrs::{
+    crypto::PublicKey,
+    proto::{
+        cosmos::{
+            auth::v1beta1::QueryAccountRequest,
+            bank::v1beta1::{QueryBalanceRequest, QueryBalanceResponse},
+            base::abci::v1beta1::TxResponse,
+        },
+        tendermint::types::Block,
+    },
+    tx::{SequenceNumber, SignerInfo, SignerPublicKey},
+    AccountId, Any, Coin, Tx,
+};
+use derive_new::new;
+
+use super::{rest::RestProvider, RpcProvider};
+use crate::{
+    ConnectionConf, CosmosAccountId, CosmosAddress, CosmosAmount, HyperlaneCosmosError, Signer,
+};
+use hyperlane_core::{
+    h512_to_bytes,
+    rpc_clients::{BlockNumberGetter, FallbackProvider},
+    utils::{self, to_atto},
+    AccountAddressType, BlockInfo, ChainCommunicationError, ChainInfo, ChainResult,
+    ContractLocator, HyperlaneChain, HyperlaneDomain, HyperlaneMessage, HyperlaneProvider,
+    HyperlaneProviderError, LogMeta, ModuleType, RawHyperlaneMessage, TxnInfo, TxnReceiptInfo,
+    H256, H512, U256,
+};
+use itertools::Itertools;
+use prost::Message;
+use reqwest::Error;
+use serde::{de::DeserializeOwned, Deserialize};
+use tendermint::{hash::Algorithm, Hash};
+use tendermint_rpc::{
+    client::CompatMode,
+    endpoint::{block, block_results, tx},
+    Client, HttpClient,
+};
+use time::OffsetDateTime;
+use tonic::async_trait;
+use tracing::{debug, trace, warn};


It is worth to do everywhere.

It will be great to format imports as follows.

std (and empty line)

external crates (and empty line)

internal crates (like hyperlane_core) (and empty line)

crate:: imports (and empty line)

super:: imports (and empty line)

Suggested change

use std::{io::Cursor, ops::Deref};

use cosmrs::{

crypto::PublicKey,

proto::{

cosmos::{

auth::v1beta1::QueryAccountRequest,

bank::v1beta1::{QueryBalanceRequest, QueryBalanceResponse},

base::abci::v1beta1::TxResponse,

},

tendermint::types::Block,

},

tx::{SequenceNumber, SignerInfo, SignerPublicKey},

AccountId, Any, Coin, Tx,

};

use derive_new::new;

use super::{rest::RestProvider, RpcProvider};

use crate::{

ConnectionConf, CosmosAccountId, CosmosAddress, CosmosAmount, HyperlaneCosmosError, Signer,

};

use hyperlane_core::{

h512_to_bytes,

rpc_clients::{BlockNumberGetter, FallbackProvider},

utils::{self, to_atto},

AccountAddressType, BlockInfo, ChainCommunicationError, ChainInfo, ChainResult,

ContractLocator, HyperlaneChain, HyperlaneDomain, HyperlaneMessage, HyperlaneProvider,

HyperlaneProviderError, LogMeta, ModuleType, RawHyperlaneMessage, TxnInfo, TxnReceiptInfo,

H256, H512, U256,

};

use itertools::Itertools;

use prost::Message;

use reqwest::Error;

use serde::{de::DeserializeOwned, Deserialize};

use tendermint::{hash::Algorithm, Hash};

use tendermint_rpc::{

client::CompatMode,

endpoint::{block, block_results, tx},

Client, HttpClient,

};

use time::OffsetDateTime;

use tonic::async_trait;

use tracing::{debug, trace, warn};

use std::{io::Cursor, ops::Deref};

use cosmrs::{

crypto::PublicKey,

proto::{

cosmos::{

auth::v1beta1::QueryAccountRequest,

bank::v1beta1::{QueryBalanceRequest, QueryBalanceResponse},

base::abci::v1beta1::TxResponse,

},

tendermint::types::Block,

},

tx::{SequenceNumber, SignerInfo, SignerPublicKey},

AccountId, Any, Coin, Tx,

};

use derive_new::new;

use itertools::Itertools;

use prost::Message;

use reqwest::Error;

use serde::{de::DeserializeOwned, Deserialize};

use tendermint::{hash::Algorithm, Hash};

use tendermint_rpc::{

client::CompatMode,

endpoint::{block, block_results, tx},

Client, HttpClient,

};

use time::OffsetDateTime;

use tonic::async_trait;

use tracing::{debug, trace, warn};

use hyperlane_core::{

h512_to_bytes,

rpc_clients::{BlockNumberGetter, FallbackProvider},

utils::{self, to_atto},

AccountAddressType, BlockInfo, ChainCommunicationError, ChainInfo, ChainResult,

ContractLocator, HyperlaneChain, HyperlaneDomain, HyperlaneMessage, HyperlaneProvider,

HyperlaneProviderError, LogMeta, ModuleType, RawHyperlaneMessage, TxnInfo, TxnReceiptInfo,

H256, H512, U256,

};

use crate::{

ConnectionConf, CosmosAccountId, CosmosAddress, CosmosAmount, HyperlaneCosmosError, Signer,

};

use super::{rest::RestProvider, RpcProvider};

Good to know, will adapt this convention 👍

ameten · 2025-02-26T16:27:24Z

I would agree with @tkporter in this comment that reducing the duplication of the code will be great since it will make the review of the change much easier.

It will be ideal to have the refactoring of the re-used parts as a separate PR and, if it too hard, at least as a separate commit so that it can be reviewed without other CosmosNative changes.

yjamin added 5 commits January 24, 2025 10:58

feat: cosmos native

66bc49e

Merge branch 'main' into feat/cosmos-native

221305b

feat: removed grpc, code cleanup

a3b5412

update: E2E Tests

8cc2a91

Merge remote-tracking branch 'fork/main' into feat/cosmos-native

48e1bf4

yjamin requested review from tkporter and daniel-savu as code owners February 4, 2025 16:54

yjamin changed the title ~~Feat: Cosmos SDK Module~~ feat: Cosmos SDK Module Feb 4, 2025

fix: typos

8670421

tkporter reviewed Feb 26, 2025

View reviewed changes

ameten reviewed Feb 26, 2025

View reviewed changes

yjamin added 8 commits March 6, 2025 10:43

chore: updated queries to adjust for the latest module changes

f2942ec

chore: import ordering

cab053c

fix: accidental url change

3547c35

fix: revert changes in chckpoint_syncer

bc794cd

fix: add error for unsupported traits instead of todo

5fa94b4

fix: revert changes in backward cursor

97a0e42

fix: revert changes in forward cursor

284c082

style: Cargo.toml formatting

3f3222a

Merge remote-tracking branch 'fork/main' into feat/cosmos-native

daf2b1e

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: Cosmos SDK Module #5370

feat: Cosmos SDK Module #5370

yjamin commented Feb 4, 2025 •

edited

Loading

changeset-bot bot commented Feb 4, 2025 •

edited

Loading

tkporter left a comment

tkporter Feb 9, 2025

yjamin Mar 5, 2025

tkporter Feb 9, 2025

tkporter Feb 9, 2025

tkporter Feb 9, 2025

tkporter Feb 9, 2025

tkporter Feb 9, 2025

ameten Feb 26, 2025

yjamin Mar 5, 2025

ameten Feb 26, 2025

ameten Feb 26, 2025

yjamin Mar 5, 2025

ameten Feb 26, 2025

yjamin Mar 5, 2025

ameten Feb 26, 2025

yjamin Mar 5, 2025

ameten Feb 26, 2025

yjamin Mar 5, 2025

ameten Feb 26, 2025

yjamin Mar 5, 2025

ameten Feb 26, 2025

yjamin Mar 5, 2025

ameten Feb 26, 2025

ameten Feb 26, 2025

yjamin Mar 5, 2025

ameten commented Feb 26, 2025

	hyperlane-cosmwasm-interface.workspace = true
	hyperlane-cosmwasm-interface = { workspace = true }

		// always write the 0th index
		if index > curr \|\| index == 0 {

feat: Cosmos SDK Module #5370

Are you sure you want to change the base?

feat: Cosmos SDK Module #5370

Conversation

yjamin commented Feb 4, 2025 • edited Loading

Description

Drive-by Fixes

Related Issues

Backward Compatibility

Testing

E2E Testing Details

changeset-bot bot commented Feb 4, 2025 • edited Loading

⚠️ No Changeset found

tkporter left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ameten commented Feb 26, 2025

yjamin commented Feb 4, 2025 •

edited

Loading

changeset-bot bot commented Feb 4, 2025 •

edited

Loading