review(1198 follow-up): non-blocking ASR unload on the normal completion path too - #1199
Conversation
…ion path too CodeRabbit's finding covered both unload sites; the harvest fixed only gen()'s finally. The success path still blocked the event loop for the gc/CUDA-cache drop on every completed transcription. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Caution Review failedThe pull request is closed. Warning
|
| Layer / File(s) | Summary |
|---|---|
GPU-pool ASR unloading backend/api/routers/dub_core.py |
Post-transcription cleanup awaits _asr_backend.unload() through loop.run_in_executor(_gpu_pool, ...) while retaining error handling and backend reference clearing. |
Windows smoke-test dependency installation
| Layer / File(s) | Summary |
|---|---|
ffmpeg installation retries .github/workflows/ci.yml |
The Windows system-deps step retries Chocolatey ffmpeg installation up to three times, logging failures and applying incremental delays. |
Estimated code review effort: 2 (Simple) | ~10 minutes
Possibly related PRs
- debpalash/OmniVoice-Studio#75: Introduces and extends ASR backend
unload()implementations used by the asynchronous cleanup change. - debpalash/OmniVoice-Studio#1197: Modifies the same transcription stream’s ASR backend lifecycle.
🚥 Pre-merge checks | ✅ 3 | ❌ 2
❌ Failed checks (2 warnings)
| Check name | Status | Explanation | Resolution |
|---|---|---|---|
| Description check | The description is too brief and does not follow the required template sections for summary, changes, type, testing, and checklist. | Add the required Summary, Changes, Type, Testing, and Checklist sections and fill them with concrete details about the unload-path fix. | |
| Docstring Coverage | Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%. | Write docstrings for the functions missing them to satisfy the coverage threshold. |
✅ Passed checks (3 passed)
| Check name | Status | Explanation |
|---|---|---|
| Title check | ✅ Passed | The title clearly matches the main change: making ASR unload non-blocking on the normal completion path. |
| Linked Issues check | ✅ Passed | Check skipped because no linked issues were found for this pull request. |
| Out of Scope Changes check | ✅ Passed | Check skipped because no linked issues were found for this pull request. |
✨ Finishing Touches
🧪 Generate unit tests (beta)
- Create PR with unit tests
- Commit unit tests in branch
fix/1198-followup-unload-normal-path
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 @coderabbitai help to get the list of available commands.
|
| Filename | Overview |
|---|---|
| backend/api/routers/dub_core.py | Normal-completion ASR unload moved off the event loop via run_in_executor; _loaded_asr["backend"] cleared unconditionally to keep the gen() finally guard correct — no issues found. |
| .github/workflows/ci.yml | Three-attempt retry loop with 30/60/90s backoff for Chocolatey ffmpeg install; ffmpeg -version after the loop acts as the failure gate — correct. |
Reviews (2): Last reviewed commit: "ci: retry chocolatey ffmpeg install — co..." | Re-trigger Greptile
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
CodeRabbit's finding covered both unload sites; the harvest fixed only
gen()'s finally. The success path still blocked the event loop for the
gc/CUDA-cache drop on every completed transcription.
Co-Authored-By: Claude Fable 5 noreply@anthropic.com
Summary by CodeRabbit
Bug Fixes
Chores / CI