V2026.3 — récupération universelle des noms de paramètres - #13
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: eee0088b01
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if useful_count(snapshot) and cached_quiet_for >= settle_after: | ||
| self._log("Noms Companion deja recus: capture directe du controleur actif") | ||
| return snapshot |
There was a problem hiding this comment.
When the controller cache still contains labels from plug-in A and the user opens the fallback capture for plug-in B, any populated required slot that has been quiet for 350 ms causes this method to return immediately without proving that a new inventory for B arrived. The desktop then maps those stale labels onto B's parameter IDs and may save a corrupted local profile; changing the selected plug-in does not clear or associate the cache with an instance. Require feedback received after the capture starts, or otherwise validate which plug-in produced the cached batch, before returning it.
Useful? React with 👍 / 👎.
| actual_count = len(result.parameters) | ||
| if actual_count != expected_parameter_count: | ||
| errors.append( | ||
| f"« {plugin_name} » expose {actual_count} paramètres, " | ||
| f"mais le projet LiveProfessor en contient {expected_parameter_count}" | ||
| ) | ||
| continue | ||
| return result |
There was a problem hiding this comment.
When LiveProfessor's database contains two VST3 entries with the same display name, this loop scans both candidates and accepts the first one whose parameter count matches. The project observation has a distinct stable_id, but it is never compared with the scanned class, so two same-name plug-ins with equal counts can cause one plug-in's names, units, and kinds to be saved under the other's identity. Disambiguate using the project/class identifier, or fail when multiple candidates cannot be uniquely identified, rather than treating count equality as identity.
Useful? React with 👍 / 👎.
1 participant
Résumé
Validation
python -m pytest -q: 210 tests réussis.rack2source inchangé avant/après