fix(strategy29): preserve fatal local monitor diagnostics - #276
Conversation
|
@codex review |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![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: 5950e8e1eb
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".
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: c4ccf5038a
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".
|
@codex review |
|
Codex Review: Didn't find any major issues. You're on a roll. Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
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". |
1 participant
Local Bollinger monitoring can enter its existing fatal state and clear its marker layer while the outer runtime remains healthy. Version 0.2.1 loses the local failure cause after cleanup, leaving source inspection unable to identify the failing chart operation.
Retain one bounded, read-only
lastLocalFailurewith the export/detect/reconcile/render stage, thrown value type, available name/message, chart context and pre-cleanup counts. The record survives cleanup and stop. String and non-Error host rejections are represented without serializing arbitrary objects. Each name/message field is read once within a narrow diagnostic boundary; throwing accessors are identified inunreadableFields. Unclassifiable host Proxy rejections explicitly stop the monitor and record classificationFailed. Normal fatal classification, snapshot-race recovery, marker rendering and the frozen detector/core remain unchanged. Publish the generated observer as 0.2.2.Validation: 824 unit/DOM tests passed; all 64 browser fixture tests passed before the rejection serialization corrections and the two affected generated coexistence cases passed again on the final artifact; Binance userscript syntax checks passed; the real Python gateway/generated JavaScript contract verifier passed for 0.2.2, including latest sequences [2, 3], following increment [4] and HTTP 400/401/409/503. Source and generated metadata match, with unchanged update URLs.
This is diagnostic instrumentation, not a claim that the intermittent local chart failure is fixed. Production Strategy29 activation and notifications remain outside this change. Exact installed-source and live diagnostics checks follow publication.