feat(sdk): add retry-with-backoff for transient localization failures#2105
Merged
Conversation
Contributor
📝 WalkthroughWalkthroughThe PR adds configurable exponential backoff retry logic to the SDK engine for transient localization failures. New ChangesLocalization Retry Resilience
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
![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
🧹 Nitpick comments (1)
packages/sdk/src/index.spec.ts (1)
605-624: ⚡ Quick winAdd a regression test for abort occurring during retry backoff.
Current coverage checks already-aborted signals, but not cancellation that happens after the first failed attempt while waiting to retry. That case would guard the new backoff sleep path directly.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@packages/sdk/src/index.spec.ts` around lines 605 - 624, Add a regression test that simulates abort during retry backoff by exercising LingoDotDevEngine.localizeObject: configure engine with retryDelayMs > 0, have the first fetch attempt fail (mockFetch rejects), start the localization call, then trigger controller.abort() while the engine is awaiting the backoff sleep (use fake timers to advance time appropriately); assert the call rejects with an "aborted" error and that mockFetch was not called a second time. This targets the retry/backoff sleep path to ensure cancellation during backoff stops further retries.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@packages/sdk/src/index.ts`:
- Around line 97-112: The sleep function can miss an abort that happens between
the early signal.aborted check and addEventListener; fix it by registering the
onAbort listener before checking signal.aborted, then immediately check
signal.aborted and if set call onAbort (or reject) to avoid the race; ensure you
still clear the timeout and remove the listener in both the timer callback and
the abort handler for proper cleanup (references: sleep, onAbort, timer).
---
Nitpick comments:
In `@packages/sdk/src/index.spec.ts`:
- Around line 605-624: Add a regression test that simulates abort during retry
backoff by exercising LingoDotDevEngine.localizeObject: configure engine with
retryDelayMs > 0, have the first fetch attempt fail (mockFetch rejects), start
the localization call, then trigger controller.abort() while the engine is
awaiting the backoff sleep (use fake timers to advance time appropriately);
assert the call rejects with an "aborted" error and that mockFetch was not
called a second time. This targets the retry/backoff sleep path to ensure
cancellation during backoff stops further retries.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 4f23073a-203d-4da3-aeff-f8659b26ecf3
📒 Files selected for processing (3)
.changeset/yummy-snails-return.mdpackages/sdk/src/index.spec.tspackages/sdk/src/index.ts
Comment on lines
+97
to
+112
| private static sleep(ms: number, signal?: AbortSignal): Promise<void> { | ||
| return new Promise((resolve, reject) => { | ||
| if (signal?.aborted) { | ||
| reject(new Error("Operation was aborted")); | ||
| return; | ||
| } | ||
| const onAbort = () => { | ||
| clearTimeout(timer); | ||
| reject(new Error("Operation was aborted")); | ||
| }; | ||
| const timer = setTimeout(() => { | ||
| signal?.removeEventListener("abort", onAbort); | ||
| resolve(); | ||
| }, ms); | ||
| signal?.addEventListener("abort", onAbort, { once: true }); | ||
| }); |
Contributor
There was a problem hiding this comment.
Fix abort-listener registration race in sleep.
An abort between the early signal.aborted check and addEventListener can be missed, causing delayed cancellation until the timeout elapses.
💡 Suggested patch
private static sleep(ms: number, signal?: AbortSignal): Promise<void> {
return new Promise((resolve, reject) => {
if (signal?.aborted) {
reject(new Error("Operation was aborted"));
return;
}
- const onAbort = () => {
- clearTimeout(timer);
- reject(new Error("Operation was aborted"));
- };
- const timer = setTimeout(() => {
- signal?.removeEventListener("abort", onAbort);
- resolve();
- }, ms);
- signal?.addEventListener("abort", onAbort, { once: true });
+ let timer: ReturnType<typeof setTimeout> | undefined;
+ const onAbort = () => {
+ if (timer) clearTimeout(timer);
+ signal?.removeEventListener("abort", onAbort);
+ reject(new Error("Operation was aborted"));
+ };
+ signal?.addEventListener("abort", onAbort, { once: true });
+ if (signal?.aborted) {
+ onAbort();
+ return;
+ }
+ timer = setTimeout(() => {
+ signal?.removeEventListener("abort", onAbort);
+ resolve();
+ }, ms);
});
}🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@packages/sdk/src/index.ts` around lines 97 - 112, The sleep function can miss
an abort that happens between the early signal.aborted check and
addEventListener; fix it by registering the onAbort listener before checking
signal.aborted, then immediately check signal.aborted and if set call onAbort
(or reject) to avoid the race; ensure you still clear the timeout and remove the
listener in both the timer callback and the abort handler for proper cleanup
(references: sleep, onAbort, timer).
cherkanovart
approved these changes
Jun 9, 2026
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
2 participants
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.
Summary
Add automatic retry-with-backoff for transient localization failures in the SDK, so a 5xx response or network blip no longer fails the whole request.
Changes
fetchWithRetrytoLingoDotDevEngine, used bylocalizeChunk: retries on 5xx responses and network errors with exponential backoff + full jitter.maxRetries(default3) andretryDelayMs(default500);maxRetries: 0disables retries.>= 500); 4xx responses and aborted requests (signal.aborted) are never retried.Testing
Business logic tests added:
maxRetrieson persistent 5xx (asserts initial attempt + N retries)Invalid request)fetch) and recoversmaxRetries: 0performs a single attempt with no retriesVisuals
Required for UI/UX changes:
N/A
Checklist
Closes N/A
Summary by CodeRabbit
New Features
maxRetries(default: 3) andretryDelayMs(default: 500ms) to control retry behavior.Tests