[rpc-alt] Use gRPC instead of JSON-RPC for Write API in sui-indexer-alt-jsonrpc by emmazzz · Pull Request #25860 · MystenLabs/sui

emmazzz · 2026-03-17T07:02:40Z

Description

Replace the JSON-RPC HTTP client delegation with gRPC calls for executeTransactionBlock and dryRunTransactionBlock. The Write module now uses FullnodeClient from sui-indexer-alt-reader, which communicates with the fullnode via TransactionExecutionService gRPC.

Add configurable read_mask parameter to FullnodeClient's execute_transaction and simulate_transaction methods
Add --fullnode-grpc-url flag (separate from existing --fullnode-rpc-url which remains for DelegationGovernance JSON-RPC delegation)
Implement proto-to-JSON-RPC response conversion for effects, events, balance changes, and object changes
Build dynamic FieldMask based on SuiTransactionBlockResponseOptions

Test plan

How did you test the new or updated feature?

Release notes

Check each box that your changes affect. If none of the boxes relate to your changes, release notes aren't required.

For each box you select, include information after the relevant heading that describes the impact of your changes that a user might notice and any actions they must take to implement updates.

Replace the JSON-RPC HTTP client delegation with gRPC calls for executeTransactionBlock and dryRunTransactionBlock. The Write module now uses FullnodeClient from sui-indexer-alt-reader, which communicates with the fullnode via TransactionExecutionService gRPC. - Add configurable read_mask parameter to FullnodeClient's execute_transaction and simulate_transaction methods - Add --fullnode-grpc-url flag (separate from existing --fullnode-rpc-url which remains for DelegationGovernance JSON-RPC delegation) - Implement proto-to-JSON-RPC response conversion for effects, events, balance changes, and object changes - Build dynamic FieldMask based on SuiTransactionBlockResponseOptions Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

vercel · 2026-03-17T07:02:45Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
sui-docs	Ready	Preview, Comment	Mar 17, 2026 6:03pm

2 Skipped Deployments

Project	Deployment	Actions	Updated (UTC)
multisig-toolkit	Ignored	Preview	Mar 17, 2026 6:03pm
sui-kiosk	Ignored	Preview	Mar 17, 2026 6:03pm

Move executeTransactionBlock and dryRunTransactionBlock tests from jsonrpc_fn_delegation_tests.rs into a new jsonrpc_write_api_tests.rs. The delegation test file now only tests governance APIs (getStakes, getStakesByIds, getValidatorsApy) and no longer needs a gRPC URL. The new write tests verify conversion correctness: sender identity in transaction input, exact balance change amounts for sender/recipient, effects status and gas costs, object change types, and consistency between dry run and execute results. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

wlmyng

i think the write.rs using grpc ends up sharing some overlapping logic with transactions/response.rs. Could extract the shared transformation into a common function. Not sure if that needs to be blocking. @amnn over to you

wlmyng · 2026-03-20T21:13:02Z

crates/sui-indexer-alt-e2e-tests/tests/jsonrpc_write_api_tests.rs

+async fn test_dry_run_invalid_tx() {
+    telemetry_subscribers::init_for_testing();
+    let cluster = WriteTestCluster::new().await.unwrap();
+
+    let response = cluster
+        .execute_jsonrpc(
+            "sui_dryRunTransactionBlock",
+            json!({ "tx_bytes": "invalid_tx_bytes" }),
+        )
+        .await
+        .unwrap();


this is a nicer error than, "Invalid value was given to the function". makes me think though, are these technically breaking changes since the error string is different?

wlmyng · 2026-03-20T21:19:28Z

crates/sui-indexer-alt-e2e-tests/tests/jsonrpc_write_api_tests.rs

+    // The transaction either fails at execution (effects with failure status)
+    // or is rejected by the validator during simulation (error response).
+    let has_failure_effects = response["result"]["effects"]["status"]["status"] == "failure";
+    let has_error = response["error"].is_object();
+    assert!(
+        has_failure_effects || has_error,
+        "expected failure effects or error response, got: {response}"
+    );
+}


is the response["error"] path just to handle transient errors? wouldn't we expect this to always have failure effects - it's a valid txn but gets aborted in the vm because not run by validator?

wlmyng · 2026-03-20T21:19:59Z

crates/sui-indexer-alt-e2e-tests/tests/jsonrpc_write_api_tests.rs

+        .await
+        .unwrap();
+
+    assert!(response["error"]["code"].is_number());


any reason we aren't asserting on the exact error cod enumber?

wlmyng · 2026-03-20T22:58:48Z

crates/sui-indexer-alt-jsonrpc/src/api/write.rs

+                result.raw_effects = effects_bcs.value().to_vec();
+            }
+
+            let effects: TransactionEffects = effects_bcs


i think we nly need to deserialize effects_bcs if show_effects or show_object_changes

wlmyng · 2026-03-20T23:38:54Z

crates/sui-indexer-alt-jsonrpc/src/api/write.rs

+}
+
+fn build_object_changes(
+    tx_data: &TransactionData,


cant we just pass the sender SuiAddress in instead of the full TransactionData?

wlmyng · 2026-03-20T23:43:50Z

crates/sui-indexer-alt-jsonrpc/src/api/write.rs

+    let mut objects: HashMap<(ObjectID, u64), Object> = HashMap::new();
+    if let Some(object_set) = &executed_tx.objects {
+        for proto_obj in &object_set.objects {
+            if let Some(bcs) = &proto_obj.bcs {


when might an obj not have a bcs? when it's Wrapped or Deleted only? Given its a proto field could we be masking when the bcs should be present but isn't due to malformed data?

wlmyng · 2026-03-20T23:46:37Z

crates/sui-indexer-alt-jsonrpc/src/api/write.rs

+                o.version().value(),
+            ),
+
+            (ID::Created, _, None) => continue,


obj created and immediately wrapped?

wlmyng · 2026-03-20T23:47:03Z

crates/sui-indexer-alt-jsonrpc/src/api/write.rs

+            ),
+
+            (ID::Created, _, None) => continue,
+            (ID::None, None, _) => continue,


what scenario might result in this? If we have an unwrap and an output object it seems we'd skip it?

wlmyng · 2026-03-20T23:51:49Z

crates/sui-indexer-alt-jsonrpc/src/api/write.rs

+            (ID::None, _, Some((o, _))) if o.is_package() => continue,
+            (ID::Deleted, None, _) => continue,


this is a package upgrade?

wlmyng · 2026-03-21T00:04:11Z

crates/sui-indexer-alt-jsonrpc/src/api/write.rs

+    Ok(SuiTransactionBlockEvents { data: sui_events })
+}
+
+fn build_object_changes(


wait a moment, is it just me or does this look really similar to the object_Changes in transactions/response.rs? could we abstract and share the logic mapping the existence of an object in input, output, and the IDOperation?

amnn · 2026-03-23T18:20:40Z

crates/sui-indexer-alt-graphql/src/api/mutation.rs

If there's only one real use case of execute_transaction and simulate_transaction that rely on field masks being set how they are, rather than using the current field mask values as defaults, can we make it a non-optional parameter and have these call-sites be explicit about what they expect?

Maybe there's a potential to consolidate the field masks that we use in GraphQL between execution and simulation as a result.

amnn · 2026-03-23T18:27:50Z

crates/sui-indexer-alt-jsonrpc/src/api/write.rs

Please take a look at the structure used for producing a transaction response for sui_getTransactionBlock (where we have a separate response module and each response option has its own associated helper function), and replicate that here.

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env March 17, 2026 07:02 — with GitHub Actions Inactive

emmazzz force-pushed the emma/use-grpc-in-rpc-alt branch from 1c942d6 to bab5c3e Compare March 17, 2026 18:01

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env March 17, 2026 18:01 — with GitHub Actions Inactive

vercel bot deployed to Preview – sui-docs March 17, 2026 18:03 View deployment

emmazzz marked this pull request as ready for review March 19, 2026 16:56

emmazzz requested a review from a team as a code owner March 19, 2026 16:56

emmazzz requested review from amnn, evan-wall-mysten, henryachen, nickvikeras, tpham-mysten and wlmyng March 19, 2026 16:56

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env March 19, 2026 16:56 — with GitHub Actions Inactive

wlmyng reviewed Mar 21, 2026

View reviewed changes

amnn reviewed Mar 23, 2026

View reviewed changes

		(ID::None, _, Some((o, _))) if o.is_package() => continue,
		(ID::Deleted, None, _) => continue,

Conversation

emmazzz commented Mar 17, 2026

Description

Test plan

Release notes

Uh oh!

vercel bot commented Mar 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

wlmyng 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!

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!

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!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

vercel bot commented Mar 17, 2026 •

edited

Loading