feat: forester: grpc processing for v2 trees #2024
Conversation
|
Note Currently processing new changes in this PR. This may take a few minutes, please wait... ⛔ Files ignored due to path filters (1)
📒 Files selected for processing (41)
WalkthroughThis PR integrates Photon gRPC service connectivity into Forester for event-driven queue updates. A new WorkCoordinator component connects to Photon, subscribes to queue updates, and distributes them to registered trees via per-tree channels. Configuration propagates the Photon gRPC URL through CLI and config layers. Proto definitions define the QueueService API. Supporting changes include v2 batch processing refactoring with parallel proof generation, addition of batch_start_index fields, and type conversions for queue indices. Changes
Sequence Diagram(s)sequenceDiagram
participant EM as EpochManager
participant WC as WorkCoordinator
participant Photon as Photon gRPC
participant Tree as Tree Task
rect rgb(200, 220, 255)
Note over EM,Photon: Initialization (photon_grpc_url present)
EM->>WC: new(photon_grpc_url)
WC->>Photon: Establish gRPC connection
Photon-->>WC: Channel ready
EM->>WC: run_dispatcher()
WC->>WC: spawn async dispatcher loop
end
rect rgb(200, 220, 255)
Note over WC,Tree: Event-driven Processing
WC->>Photon: SubscribeQueueUpdates
Photon-->>WC: stream QueueUpdate
WC->>WC: register_tree(tree_pubkey)
WC-->>Tree: mpsc::Receiver<QueueUpdateMessage>
Photon-->>WC: Queue update event
WC->>Tree: Send update via channel
Tree->>Tree: Process event-driven work
end
rect rgb(255, 220, 200)
Note over WC,Photon: Fallback on Disconnect
WC->>Photon: Reconnect attempt (exponential backoff)
alt Reconnect succeeds
Photon-->>WC: New connection
else Max retries exceeded
WC->>WC: Fall back to polling
Note over WC: EpochManager switches to polling mode
end
end
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~30 minutes
Possibly related PRs
Suggested reviewers
Poem
Pre-merge checks and finishing touches❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
sergeytimoshin
changed the title
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 12
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
cli/src/utils/processPhotonIndexer.ts (1)
55-62: Fix incorrect gRPC flag and value type; use--grpc-urlwith full URL endpointThe code uses wrong flag name and value type. Photon CLI uses
--grpc-url(not--grpc-port) to configure gRPC endpoint. The current"--grpc-port", "50051"passes a bare port instead of a URL, causing the flag to be unrecognized or ignored.Required changes:
- Flag:
--grpc-port→--grpc-url- Value:
"50051"→ full URL like"http://localhost:50051"- Add optional
grpcUrlparameter to function signatureSuggested fix:
+export const PHOTON_GRPC_URL = "http://localhost:50051"; export async function startIndexer( rpcUrl: string, indexerPort: number, checkPhotonVersion: boolean = true, photonDatabaseUrl?: string, + grpcUrl: string = PHOTON_GRPC_URL, ) { ... const args: string[] = [ "--port", indexerPort.toString(), "--rpc-url", rpcUrl, - "--grpc-port", - "50051", + "--grpc-url", + grpcUrl, ];
🧹 Nitpick comments (15)
forester/src/processor/v1/helpers.rs (1)
172-260: Consider removing or making the debug code configurable.This debug block is currently dead code since
_debugis hardcoded tofalse. While the diagnostic logic for validating on-chain state and root history could be valuable for troubleshooting, it adds ~90 lines of unmaintained code with no activation mechanism.Consider one of the following:
- Remove the debug block if it's no longer needed for active development.
- Make it configurable via an environment variable or feature flag so it can be enabled without source modifications:
- let _debug = false; - if _debug { + if std::env::var("FORESTER_DEBUG_STATE_PROOFS").is_ok() {Additionally, this PR focuses on v2 tree processing according to the title, but these changes are in the v1 helpers. Please clarify if this debug code is needed for v2 work or if it was left in accidentally.
forester-utils/Cargo.toml (1)
51-53: Dev deps look fine; consider constraining features/pinning via workspaceTo keep compile surface minimal and versions consistent:
- Prefer default-features = false on tokio-postgres unless TLS is needed via companion crates.
- Pin bs58 in the workspace and reference a version here (already using workspace = true).
Example:
[dev-dependencies] tokio-postgres = { version = "0.7", default-features = false } bs58 = { workspace = true }forester/src/forester_status.rs (2)
198-203: Use centralized skip logic (should_skip_tree) for consistencyPrefer the new helper to encapsulate skip rules and avoid drift if criteria expand.
For example:
- // Skip rolled-over trees - if tree.is_rolledover { + // Skip trees not eligible for processing + if crate::epoch_manager::should_skip_tree(tree) { continue; }Confirm run_queue_info earlier doesn’t need the same filter to keep metrics and status aligned.
256-259: Build active_trees via helper to avoid duplicating criteriaConstruct the filtered list using should_skip_tree to ensure the same rules apply everywhere.
- // Filter out rolled-over trees - let active_trees: Vec<TreeAccounts> = - trees.iter().filter(|t| !t.is_rolledover).cloned().collect(); + // Filter out ineligible trees + let active_trees: Vec<TreeAccounts> = + trees.iter().filter(|t| !crate::epoch_manager::should_skip_tree(t)).cloned().collect();forester/src/slot_tracker.rs (1)
86-113: Harden against transient RPC errors and consider parameterizing MAX_SLEEP_SLOTSThe loop currently bails on any get_slot() error; brief RPC blips will error out the caller.
- Wrap get_slot() in a small retry/backoff and continue on error.
- Make MAX_SLEEP_SLOTS configurable (env/arg) so tests and different slot times can tune it.
- Optional: compute sleep without floats:
- const MAX_SLEEP_SLOTS: u64 = 50; // ~20 seconds max sleep between checks + const MAX_SLEEP_SLOTS: u64 = 50; // ~20–23s depending on slot duration @@ - let sleep_duration = - Duration::from_secs_f64(sleep_slots as f64 * slot_duration().as_secs_f64()); + let per_slot = slot_duration(); + let sleep_duration = per_slot.saturating_mul(sleep_slots as u32);Where saturating_mul is available from Duration on recent Rust; otherwise multiply nanos with checked math.
If keeping errors fatal is intentional (e.g., upstream retries), please confirm.
forester-utils/src/instructions/address_batch_update.rs (3)
146-159: Bound parallel proof generation; preserve output orderjoin_all fires all requests at once; large chunks can overwhelm the prover and memory.
- Limit concurrency with a buffer (e.g., PROVER_MAX_CONCURRENCY).
- Preserve order by indexing results and emitting in-order.
Illustrative sketch:
- let proof_futures: Vec<_> = all_inputs.into_iter().enumerate().map(|(i, inputs)| { - let client = Arc::clone(&proof_client); - async move { - let result = client.generate_batch_address_append_proof(inputs).await; - (i, result) - } - }).collect(); - - // Wait for all proofs to complete in parallel - let proof_results = futures::future::join_all(proof_futures).await; + let max_conc = std::cmp::max(1, std::env::var("PROVER_MAX_CONCURRENCY").ok() + .and_then(|v| v.parse::<usize>().ok()).unwrap_or(8)); + let proof_results = futures::stream::iter( + all_inputs.into_iter().enumerate().map(|(i, inputs)| { + let client = Arc::clone(&proof_client); + async move { (i, client.generate_batch_address_append_proof(inputs).await) } + }) + ) + .buffer_unordered(max_conc) + .collect::<Vec<_>>() + .await; + let mut proof_results = { + let mut v = proof_results; + v.sort_by_key(|(i, _)| *i); + v + };This keeps memory bounded and preserves the intended proof order.
175-179: Avoid cloning proof_buffer when yieldingClone allocates and copies every time; use mem::take to move out efficiently.
- if proof_buffer.len() >= MAX_PROOFS_PER_TX { - yield Ok(proof_buffer.clone()); - proof_buffer.clear(); - } + if proof_buffer.len() >= MAX_PROOFS_PER_TX { + let out = std::mem::take(&mut proof_buffer); + yield Ok(out); + }
74-87: Minor: clarify param naming (queue index vs offset)Trait signature uses start_offset but we pass a queue index; name the local var start_queue_index (or add a comment) to avoid confusion across indexer backends.
- let indexer_update_info = { + let indexer_update_info = { let mut connection = rpc_pool.get_connection().await?; let indexer = connection.indexer_mut()?; debug!( "Requesting {} addresses from Photon for chunk {} with start_queue_index={:?}", elements_for_chunk, chunk_idx, next_queue_index ); match indexer .get_address_queue_with_proofs( &merkle_tree_pubkey, elements_for_chunk as u16, next_queue_index, None, )forester/src/processor/v2/common.rs (2)
522-533: Reduce debug payload and avoid leaking large arrays in logsCurrent debug logs dump entire hash chains; this is noisy and can bloat logs in active phases.
Consider logging only count + short prefixes/suffixes, or guard with a higher verbosity flag:
- e.g., map each hash chain to hex_prefix(8) and cap to first/last N entries.
- Or move detailed dumps to trace!.
516-521: Avoid potential panic on direct indexing of hash_chain_storesAccessing
hash_chain_stores[batch_index as usize][i as usize]can panic if on-chain data is inconsistent.Apply safe access with
get()and skip missing entries:- for i in num_inserted_zkps..current_zkp_batch_index { - leaves_hash_chains - .push(merkle_tree.hash_chain_stores[batch_index as usize][i as usize]); - } + if let Some(batch_store) = merkle_tree.hash_chain_stores.get(batch_index as usize) { + for i in num_inserted_zkps..current_zkp_batch_index { + if let Some(h) = batch_store.get(i as usize) { + leaves_hash_chains.push(*h); + } else { + trace!("Missing hash chain entry at i={} (batch_index={})", i, batch_index); + } + } + } else { + trace!("Missing hash_chain_stores row for batch_index={}", batch_index); + }forester/Cargo.toml (2)
56-62: Enable TLS for gRPC (if using https endpoints)If
photon_grpc_urlcan behttps://…, enable tonic TLS features or connections will fail at runtime.Apply:
-tonic = "0.14.2" +tonic = { version = "0.14.2", features = ["tls", "tls-roots"] }If Photon is strictly plaintext (http), ignore this.
56-65: Question: Do we need thetonic-prostruntime crate?Code only uses
tonicclient types; generation happens in build.rs viatonic-prost-build. Thetonic-prostruntime dep appears unused.Consider removing
tonic-prostfrom [dependencies] to avoid extra surface area unless it’s used elsewhere.forester/src/epoch_manager.rs (2)
944-967: Redundantuse_eventscheck
use_eventsequalsqueue_update_rx.is_some(), then checked again inside the V2 branch.Inline
queue_update_rx.is_some()and dropuse_eventsfor clarity.
1350-1487: Event loop is solid; ensure resilience when coordinator restartsIf the coordinator restarts and drops senders,
recv()returnsNone. Yourselect!won’t match that branch; fallback tick covers this. Consider logging once ifrecv()returnsNoneto aid debugging.forester/src/work_coordinator.rs (1)
117-126: Optional: subscribe per-iteration with a fresh clientIf you prefer to avoid interior mutability, re-create the client per loop iteration:
let mut stream = QueueServiceClient::connect(self.photon_grpc_url.clone()) .await? .subscribe_queue_updates(request) .await? .into_inner();This removes the need for
reconnect()and thegrpc_clientfield.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
⛔ Files ignored due to path filters (1)
Cargo.lockis excluded by!**/*.lock
📒 Files selected for processing (25)
cli/src/utils/processPhotonIndexer.ts(1 hunks)forester-utils/Cargo.toml(1 hunks)forester-utils/src/instructions/address_batch_update.rs(10 hunks)forester-utils/src/lib.rs(1 hunks)forester/Cargo.toml(2 hunks)forester/build.rs(1 hunks)forester/proto/photon.proto(1 hunks)forester/src/cli.rs(1 hunks)forester/src/config.rs(4 hunks)forester/src/epoch_manager.rs(20 hunks)forester/src/forester_status.rs(2 hunks)forester/src/lib.rs(2 hunks)forester/src/processor/v1/helpers.rs(1 hunks)forester/src/processor/v1/send_transaction.rs(1 hunks)forester/src/processor/v2/common.rs(2 hunks)forester/src/slot_tracker.rs(1 hunks)forester/src/tree_data_sync.rs(4 hunks)forester/src/work_coordinator.rs(1 hunks)forester/tests/e2e_test.rs(2 hunks)forester/tests/priority_fee_test.rs(1 hunks)forester/tests/test_utils.rs(1 hunks)prover/client/src/proof_types/batch_address_append/proof_inputs.rs(1 hunks)sdk-libs/client/src/indexer/photon_indexer.rs(1 hunks)sdk-libs/photon-api/src/models/_get_queue_elements_post_200_response_result.rs(1 hunks)sdk-libs/photon-api/src/models/account_context.rs(1 hunks)
🧰 Additional context used
🧠 Learnings (8)
📚 Learning: 2025-10-11T21:59:25.222Z
Learnt from: CR
PR: Lightprotocol/light-protocol#0
File: program-libs/account-checks/CLAUDE.md:0-0
Timestamp: 2025-10-11T21:59:25.222Z
Learning: Applies to program-libs/account-checks/src/**/*.rs : Validate account type with 8-byte discriminators using check_discriminator before deserialization
Applied to files:
forester/src/tree_data_sync.rs
📚 Learning: 2025-10-11T21:59:25.222Z
Learnt from: CR
PR: Lightprotocol/light-protocol#0
File: program-libs/account-checks/CLAUDE.md:0-0
Timestamp: 2025-10-11T21:59:25.222Z
Learning: Applies to program-libs/account-checks/src/discriminator.rs : Define and keep discriminator constants and the Discriminator trait in discriminator.rs for compile-time verification
Applied to files:
forester/src/tree_data_sync.rs
📚 Learning: 2025-10-11T21:59:52.712Z
Learnt from: CR
PR: Lightprotocol/light-protocol#0
File: program-libs/account-checks/docs/CLAUDE.md:0-0
Timestamp: 2025-10-11T21:59:52.712Z
Learning: Applies to program-libs/account-checks/docs/**/DISCRIMINATOR.md : DISCRIMINATOR.md must document 8-byte discriminators, the Discriminator trait, constant arrays for compile-time checks, and integration with account initialization
Applied to files:
forester/src/tree_data_sync.rs
📚 Learning: 2025-10-11T21:59:25.222Z
Learnt from: CR
PR: Lightprotocol/light-protocol#0
File: program-libs/account-checks/CLAUDE.md:0-0
Timestamp: 2025-10-11T21:59:25.222Z
Learning: Applies to program-libs/account-checks/src/**/*.rs : On account initialization, call account_info_init to set the 8-byte discriminator
Applied to files:
forester/src/tree_data_sync.rs
📚 Learning: 2025-10-11T21:59:25.222Z
Learnt from: CR
PR: Lightprotocol/light-protocol#0
File: program-libs/account-checks/CLAUDE.md:0-0
Timestamp: 2025-10-11T21:59:25.222Z
Learning: Applies to program-libs/account-checks/src/checks.rs : Expose and maintain account validation helpers (check_owner, check_program, check_mut/non_mut, check_signer, check_discriminator, set_discriminator, check_pda_seeds, check_account_balance_is_rent_exempt, account_info_init) in checks.rs
Applied to files:
forester/src/tree_data_sync.rs
📚 Learning: 2025-10-15T03:46:26.767Z
Learnt from: CR
PR: Lightprotocol/light-protocol#0
File: programs/registry/CLAUDE.md:0-0
Timestamp: 2025-10-15T03:46:26.767Z
Learning: Applies to programs/registry/src/lib.rs : Load accounts according to type before check_forester: batched via BatchedMerkleTreeAccount::type_from_account_info(); regular via ctx.accounts.account.load()?.metadata; use custom deserialization when required.
Applied to files:
forester/src/tree_data_sync.rs
📚 Learning: 2025-10-16T06:33:19.426Z
Learnt from: CR
PR: Lightprotocol/light-protocol#0
File: program-libs/compressible/CLAUDE.md:0-0
Timestamp: 2025-10-16T06:33:19.426Z
Learning: Applies to program-libs/compressible/src/config.rs : Ensure serialization compatibility across Anchor, Pinocchio, and Borsh for core account types used by dependent programs
Applied to files:
forester/src/tree_data_sync.rs
📚 Learning: 2025-10-11T21:59:25.222Z
Learnt from: CR
PR: Lightprotocol/light-protocol#0
File: program-libs/account-checks/CLAUDE.md:0-0
Timestamp: 2025-10-11T21:59:25.222Z
Learning: Applies to program-libs/account-checks/**/Cargo.toml : Define features solana, pinocchio, and test-only in Cargo.toml; default build should enable none
Applied to files:
forester/Cargo.toml
🧬 Code graph analysis (7)
forester/src/processor/v1/helpers.rs (2)
program-tests/merkle-tree/src/lib.rs (1)
root(190-195)program-libs/zero-copy/src/cyclic_vec.rs (1)
first_index(200-206)
prover/client/src/proof_types/batch_address_append/proof_inputs.rs (1)
prover/client/tests/batch_address_append.rs (1)
new_element_values(64-67)
forester-utils/src/instructions/address_batch_update.rs (7)
sdk-libs/client/src/indexer/photon_indexer.rs (2)
get_address_queue_with_proofs(1473-1579)new(116-127)sdk-libs/program-test/src/indexer/test_indexer.rs (3)
get_address_queue_with_proofs(875-959)clone(113-125)new(1309-1370)sdk-libs/client/src/indexer/indexer_trait.rs (1)
get_address_queue_with_proofs(179-185)sdk-libs/client/src/indexer/types.rs (3)
value(373-377)value(379-383)value(830-841)sdk-libs/client/src/indexer/error.rs (1)
clone(113-159)prover/client/src/proof_client.rs (1)
generate_batch_address_append_proof(525-533)program-libs/hasher/src/hash_chain.rs (1)
create_hash_chain_from_slice(23-32)
forester/src/processor/v2/common.rs (1)
sdk-libs/program-test/src/logging/config.rs (1)
debug(59-71)
forester/src/epoch_manager.rs (5)
forester/src/work_coordinator.rs (2)
new(45-60)is_healthy(192-194)forester/src/slot_tracker.rs (3)
new(25-35)wait_until_slot_reached(79-116)update(37-44)sdk-libs/client/src/rpc/rpc_trait.rs (2)
new(34-42)new(76-78)forester-utils/src/forester_epoch.rs (3)
new(68-80)new(156-161)get_epoch_phases(232-273)forester/src/config.rs (7)
clone(353-368)default(89-100)default(142-149)default(153-158)default(162-172)test_address_v2(104-115)test_state_v2(117-128)
forester/src/tree_data_sync.rs (2)
program-libs/batched-merkle-tree/src/queue.rs (1)
light_account_checks(184-184)program-libs/batched-merkle-tree/src/merkle_tree.rs (4)
light_account_checks(137-137)BatchedMerkleTreeAccount(1080-1080)BatchedMerkleTreeAccount(1090-1090)BatchedMerkleTreeAccount(1100-1100)
forester/src/work_coordinator.rs (1)
forester/src/epoch_manager.rs (2)
new(135-195)e(573-573)
🪛 Buf (1.59.0)
forester/proto/photon.proto
3-3: Files with package "photon" must be within a directory "photon" relative to root but were in directory "forester/proto".
(PACKAGE_DIRECTORY_MATCH)
🔇 Additional comments (21)
sdk-libs/photon-api/src/models/account_context.rs (1)
27-27: LGTM: Queue index widened to u64.The type change from
u16tou64fornullifier_queue_indexaligns with the v2 API migration and accommodates larger queue indices.sdk-libs/client/src/indexer/photon_indexer.rs (1)
1658-1658: LGTM: Cast removed following API type change.The direct assignment is now correct since
api_result.first_value_queue_indexisu64, matching the expected type.sdk-libs/photon-api/src/models/_get_queue_elements_post_200_response_result.rs (1)
18-18: LGTM: Queue index widened to u64.Both the field type and constructor parameter are consistently updated from
u16tou64, supporting larger queue indices in the v2 API.Also applies to: 26-26
forester/src/processor/v1/send_transaction.rs (1)
354-354: LGTM: Improved logging consistency.Elevating the log level to
error!for transaction send failures when a signature is present improves consistency with line 356 and appropriately reflects the severity when a submitted transaction fails.forester/src/cli.rs (1)
71-72: LGTM!The
photon_grpc_urlfield is correctly added following the same pattern as other optional URL fields in the struct.forester/src/lib.rs (2)
22-22: LGTM!The new
work_coordinatormodule is correctly exposed as a public module, enabling event-driven queue updates via gRPC.
66-66: LGTM!Adding
!t.is_rolledoverto the filter correctly excludes rolled-over trees from queue processing, preventing unnecessary work on inactive trees.forester/tests/test_utils.rs (1)
92-92: LGTM!Setting
photon_grpc_url: Nonein the test configuration is appropriate for tests that don't require gRPC connectivity.forester/tests/e2e_test.rs (2)
124-129: LGTM!The
get_photon_grpc_url()helper follows the established pattern of other URL helpers, correctly returning the local gRPC endpoint for local testing and reading from environment for Devnet.
212-212: LGTM!The
photon_grpc_urlfield is correctly populated from the helper function, maintaining consistency with the configuration pattern.forester/tests/priority_fee_test.rs (1)
55-55: LGTM!Setting
photon_grpc_url: Noneis appropriate for this priority fee test that doesn't require gRPC connectivity.forester/src/config.rs (4)
43-43: LGTM!The
photon_grpc_urlfield is correctly added toExternalServicesConfig, maintaining consistency with other optional service URLs.
236-236: LGTM!The
photon_grpc_urlis correctly propagated fromStartArgsintoExternalServicesConfigduring configuration initialization.
315-315: LGTM!Setting
photon_grpc_url: Noneinnew_for_statusis correct, as status checks don't require gRPC connectivity.
131-131: The Default derive concern is technically valid but not a current runtime issue.The derived
DefaultonRpcPoolConfig(line 131) would indeed produce all-zero values if called, which differs from the intentional hardcoded values (max_size: 10, connection_timeout_secs: 15, etc.). However, verification found:
- Both instantiation points (lines 280 and 336) use explicit full field initialization
- No
Default::default()calls onRpcPoolConfigexist anywhere in the codebase- No struct update syntax or implicit patterns could trigger
DefaultRpcPoolConfigis only used withinconfig.rsThe derived
Defaultis currently unused dead code. While the concern is valid and removingDefaultor implementing it properly would be defensive programming, this doesn't pose an immediate runtime risk with current usage patterns.forester/build.rs (1)
1-6: No issues found. Proto file verified to exist and build script is correctly implemented.The verification confirms that
forester/proto/photon.protoexists at the expected location with valid proto3 syntax. The build script correctly references this file with the proper paths ("proto/photon.proto"and include directory"proto"). The implementation appropriately handles dependencies with theprintln!directive forcargo:rerun-if-changed.forester-utils/src/lib.rs (1)
25-25: No action required—batch_start_indexinitialization is verified at all construction sites.The verification confirms there is exactly one construction site for
ParsedMerkleTreeDatain the codebase (forester/src/processor/v2/common.rs:534–544), and it properly initializes the newbatch_start_indexfield frombatch.start_indexat line 542. No Default implementation exists that could bypass initialization, and all other references to the struct are type annotations or imports, not construction sites.forester/src/processor/v2/common.rs (1)
542-542: LGTM: batch_start_index populatedPopulating
batch_start_indexfrombatch.start_indexcloses data-gap for downstream consumers.forester/src/epoch_manager.rs (2)
146-179: Startup coordinator wiring looks good; keep running even if initial connect failsGraceful fallback to polling with clear logs. No changes requested.
1993-2001: Nice: centralized skip logic with testsHelper + tests improve maintainability and scheduling clarity.
forester/proto/photon.proto (1)
3-3: Verify and apply Buf package-directory fixThe Buf linting rule PACKAGE_DIRECTORY_MATCH is correctly identified: package
photonatforester/proto/photon.protoviolates the convention requiring package names to match directory structure.Recommended option: Move file to
forester/proto/photon/photon.proto, keepingpackage photon;and the existinginclude_proto!("photon")call (it resolves by package name, not path). This requires no Rust code changes.Alternative: Change package declaration to
photon_protoor similar to match the current path, then update theinclude_proto!()call inforester/src/work_coordinator.rs:20accordingly.Verify your project's proto organization conventions and Buf enforcement before proceeding.
sergeytimoshin
force-pushed
the
sergey/grpc-forester
branch
from
15dcf48 to
04a9e24
Compare
sergeytimoshin
changed the title
Summary by CodeRabbit
New Features
Improvements
Observability
Tests / Chores