Make RaindexClient Send+Sync on native targets#2489
Open
findolor wants to merge 6 commits into2026-03-03-raindex-client-localdb-refactorfrom
Open
Make RaindexClient Send+Sync on native targets#2489findolor wants to merge 6 commits into2026-03-03-raindex-client-localdb-refactorfrom
findolor wants to merge 6 commits into2026-03-03-raindex-client-localdb-refactorfrom
Conversation
- Conditionally use async_trait(?Send) for WASM, async_trait for native on LocalDbQueryExecutor trait - Add FnPtr<T> alias (Rc on WASM, Arc on native) for LocalDb internals - Split LocalDb::new into WASM and native impl blocks with correct bounds - Add ClientRef type alias (Rc on WASM, Arc on native) - Gate JsCallbackExecutor module and impl to WASM-only
- Split db/scheduler fields: Rc<RefCell<...>> on WASM, Arc<Mutex<...>> on native - Split LocalDbState::new into per-target impl blocks - Use poison-safe lock access (ok()/if-let) in db(), mark_ready(), is_ready(), and Drop impl
- Replace Rc::new/Rc::clone with ClientRef::new/ClientRef::clone - Update OrdersDataSource, VaultsDataSource traits to use async_trait cfg gate - Update LocalDbOrders, LocalDbVaults to accept ClientRef - Update RaindexOrder, RaindexVault to store ClientRef
- Move clear_tables_wasm into #[cfg(target_family = "wasm")] mod wasm - Move get_sync_status into #[cfg(target_family = "wasm")] mod wasm - Keep core logic functions ungated for both targets
- Apply async_trait cfg gate to all LocalDbQueryExecutor test impls - Replace RefCell with Mutex in test executors - Replace Rc with Arc in test code
Contributor
WalkthroughThis PR introduces wasm/non-wasm specialization across the codebase. It replaces platform-agnostic synchronization primitives with conditional compilation: Rc/RefCell on single-threaded wasm and Arc/Mutex on multi-threaded native. It introduces a Changes
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes 🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches
🧪 Generate unit tests (beta)
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 |
4 tasks
Collaborator
Author
|
@coderabbitai review |
Contributor
✅ Actions performedReview triggered.
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
crates/common/src/raindex_client/local_db/mod.rs (1)
337-340:⚠️ Potential issue | 🔴 CriticalFix WASM test compile failure from missing
RaindexClientimport.Line [391] uses
RaindexClientwithout bringing it into scope, matching the current CI error.
Line [338] also has an unusedTimeoutFutureimport warning.Suggested patch
#[cfg(all(test, target_family = "wasm", feature = "browser-tests"))] mod wasm_tests { use super::*; - use gloo_timers::future::TimeoutFuture; + use crate::raindex_client::RaindexClient; use rain_orderbook_app_settings::spec_version::SpecVersion;Also applies to: 391-393
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@crates/common/src/raindex_client/local_db/mod.rs` around lines 337 - 340, Add a missing import for RaindexClient and remove the unused TimeoutFuture import: bring RaindexClient into scope (e.g., add a use statement referencing RaindexClient so the code that references RaindexClient at lines ~391–393 compiles), and delete the unused use gloo_timers::future::TimeoutFuture; line (or gate it behind a wasm-only cfg if it will be used conditionally) to silence the unused-import warning.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@crates/common/src/raindex_client/local_db/state.rs`:
- Around line 91-96: The code currently swallows poisoned Mutex locks (e.g., in
the db() accessor and the other lock sites referenced) by mapping .lock().ok()
to None/false; instead recover the inner guard from a poisoned lock so local DB
routing and cleanup remain deterministic. Replace .lock().ok().and_then(|guard|
guard.as_ref().cloned()) with lock handling that matches on self.db.lock(): on
Ok(guard) use guard.as_ref().cloned(); on Err(poisoned) call
poisoned.into_inner() and then guard.as_ref().cloned(); apply the same pattern
to the other mutex accesses (the places around the other boolean/guard checks in
the file) so that poison does not silently fall back to subgraph or skip
scheduler stop. Ensure the logic still clones/returns LocalDb (or boolean) as
before after recovering the inner guard.
---
Outside diff comments:
In `@crates/common/src/raindex_client/local_db/mod.rs`:
- Around line 337-340: Add a missing import for RaindexClient and remove the
unused TimeoutFuture import: bring RaindexClient into scope (e.g., add a use
statement referencing RaindexClient so the code that references RaindexClient at
lines ~391–393 compiles), and delete the unused use
gloo_timers::future::TimeoutFuture; line (or gate it behind a wasm-only cfg if
it will be used conditionally) to silence the unused-import warning.
ℹ️ Review info
Configuration used: Organization UI
Review profile: ASSERTIVE
Plan: Pro
📒 Files selected for processing (26)
crates/cli/src/commands/local_db/pipeline/bootstrap.rscrates/common/src/local_db/executor.rscrates/common/src/local_db/export.rscrates/common/src/local_db/pipeline/adapters/apply.rscrates/common/src/local_db/pipeline/adapters/bootstrap.rscrates/common/src/local_db/pipeline/adapters/tokens.rscrates/common/src/local_db/pipeline/adapters/window.rscrates/common/src/local_db/pipeline/engine.rscrates/common/src/local_db/query/executor.rscrates/common/src/raindex_client/add_orders.rscrates/common/src/raindex_client/local_db/executor.rscrates/common/src/raindex_client/local_db/mod.rscrates/common/src/raindex_client/local_db/orders.rscrates/common/src/raindex_client/local_db/pipeline/bootstrap.rscrates/common/src/raindex_client/local_db/pipeline/runner/mod.rscrates/common/src/raindex_client/local_db/pipeline/runner/scheduler/native.rscrates/common/src/raindex_client/local_db/query/clear_tables.rscrates/common/src/raindex_client/local_db/query/fetch_last_synced_block.rscrates/common/src/raindex_client/local_db/state.rscrates/common/src/raindex_client/local_db/vaults.rscrates/common/src/raindex_client/mod.rscrates/common/src/raindex_client/orders.rscrates/common/src/raindex_client/remove_orders.rscrates/common/src/raindex_client/trades.rscrates/common/src/raindex_client/vaults.rscrates/js_api/src/registry.rs
Comment on lines
+91
to
+96
| fn db(&self) -> Option<LocalDb> { | ||
| self.db | ||
| .lock() | ||
| .ok() | ||
| .and_then(|guard| guard.as_ref().cloned()) | ||
| } |
Contributor
There was a problem hiding this comment.
Do not silently degrade behavior on poisoned locks.
On native, lock poison currently maps to None/false/no-op (Line 93, Line 193, Line 205), which can route queries to subgraph or skip scheduler stop after a panic. Prefer recovering with into_inner() so local DB routing and cleanup remain deterministic.
Suggested fix
fn db(&self) -> Option<LocalDb> {
- self.db
- .lock()
- .ok()
- .and_then(|guard| guard.as_ref().cloned())
+ match self.db.lock() {
+ Ok(guard) => guard.as_ref().cloned(),
+ Err(poisoned) => poisoned.into_inner().as_ref().cloned(),
+ }
}
@@
- if let Ok(mut guard) = self.scheduler.lock() {
- if let Some(handle) = guard.take() {
- handle.stop();
- }
- }
+ let mut guard = self
+ .scheduler
+ .lock()
+ .unwrap_or_else(|poisoned| poisoned.into_inner());
+ if let Some(handle) = guard.take() {
+ handle.stop();
+ }
}
}
@@
- if let Ok(mut guard) = self.ready.write() {
- guard.insert(chain_id);
- }
+ let mut guard = self
+ .ready
+ .write()
+ .unwrap_or_else(|poisoned| poisoned.into_inner());
+ guard.insert(chain_id);
@@
- return self
- .ready
- .read()
- .map(|guard| guard.contains(&chain_id))
- .unwrap_or(false);
+ let guard = self
+ .ready
+ .read()
+ .unwrap_or_else(|poisoned| poisoned.into_inner());
+ return guard.contains(&chain_id);Based on learnings: when local DB is enabled and available in crates/common/src/raindex_client, it should be used exclusively without subgraph fallback, because mixing sources can cause inconsistencies.
Also applies to: 160-165, 193-206
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@crates/common/src/raindex_client/local_db/state.rs` around lines 91 - 96, The
code currently swallows poisoned Mutex locks (e.g., in the db() accessor and the
other lock sites referenced) by mapping .lock().ok() to None/false; instead
recover the inner guard from a poisoned lock so local DB routing and cleanup
remain deterministic. Replace .lock().ok().and_then(|guard|
guard.as_ref().cloned()) with lock handling that matches on self.db.lock(): on
Ok(guard) use guard.as_ref().cloned(); on Err(poisoned) call
poisoned.into_inner() and then guard.as_ref().cloned(); apply the same pattern
to the other mutex accesses (the places around the other boolean/guard checks in
the file) so that poison does not silently fall back to subgraph or skip
scheduler stop. Ensure the logic still clones/returns LocalDb (or boolean) as
before after recovering the inner guard.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Dependent PRs
Motivation
Following the local DB refactor in #2488,
RaindexClientand its dependency graph (LocalDb,LocalDbState,LocalDbQueryExecutor) still usedRc/RefCelleverywhere, making them!Send + !Syncon native targets. This blocks usingRaindexClientacross threads in CLI/server contexts.Solution
1.
LocalDbQueryExecutortrait — conditionalSendboundThe trait now uses
async_trait(?Send)on WASM andasync_trait(withSend) on native, via#[cfg_attr].2.
LocalDb—Arc-based internals on nativeFnPtr<T>alias:Rc<T>on WASM,Arc<T>on nativeExecuteBatchFn, etc.) requireSend + Syncon nativeimpl LocalDbblocks with platform-appropriate bounds3.
LocalDbState— thread-safe fields on nativedbfield:Rc<RefCell<...>>on WASM,Arc<Mutex<...>>on nativeschedulerfield: same split.ok(),if let Ok(...),.unwrap_or(false))4.
ClientReftype aliasRc<RaindexClient>on WASM,Arc<RaindexClient>on nativeorders,vaults,trades,add_orders,remove_orders)5. WASM export isolation
clear_tables_wasmandget_sync_statusmoved into#[cfg(target_family = "wasm")] mod wasmblocksJsCallbackExecutormodule gated to WASM-only6. Test updates
LocalDbQueryExecutormock impls updated withasync_traitcfg gateRefCell→Mutex,Rc→Arcin test codeChecks
By submitting this for review, I'm confirming I've done the following:
Summary by CodeRabbit
Release Notes