Conversation
Document the goal of running commands to completion (collecting issues at phase boundaries) versus aborting only on ToolkitErrors, which are by definition unrecoverable. Covers the three coexisting severity systems (legacy tk_warnings SeverityLevel, build v2 numeric InsightDefinition, dataio Severity) and the NEAT adapter that originates the Error/Warning naming mismatch in CDF-27907. Related: CDF-27907, CDF-27968.
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize the Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counterproductive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. |
Add a "Grounding in ruff" section explaining what we adopt directly (run-to-completion, stable rule codes, per-rule docs, user-driven suppression), what we adapt (categorical severity instead of ruff's binary model; ToolkitError because we have side effects), and a ruff↔toolkit concept mapping. Add stable rule codes (TK001, etc.) to the consequences list since they unlock non-breaking rename and suppression.
Merge the in-progress opener with the prior content and trim. Three categorical severities for configuration quality, ToolkitError as a fourth category outside the severity system for technical exceptions. Drops the standalone Goal section (folded into What), the long Grounding-in-ruff appendix (reduced to a single Why bullet), and condenses Why and Consequences. Replaces NOTICE / HINT with just HINT — walking through real classes showed every NOTICE candidate has the same author rule as HINT (advisory / recommendation), so the middle bucket carried no distinct signal.
…cy check The CDF API returns `subagents: []` on every AgentResponse even though the field is not in the spec and local YAMLs never define it. Add `subagents` to the existing list of undocumented empty fields that are stripped from the CDF dump when absent in the local config, matching the same handling already in place for `labels`, `exampleQuestions`, and `skills`.
☂️ Code Coverage
Overall Coverage
New FilesNo new covered files... Modified FilesNo covered modified files...
|
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #3004 +/- ##
=======================================
Coverage 85.92% 85.92%
=======================================
Files 469 469
Lines 43047 43047
=======================================
+ Hits 36986 36989 +3
+ Misses 6061 6058 -3 🚀 New features to boost your workflow:
|
3 participants
Summary
ToolkitErrorreserved for unrecoverable conditions that abort immediately.tk_warnings.SeverityLevel, build v2 numericInsightDefinition,dataio/logger.Severity) and thecognite.neat._toolkit_adaptermapping where theError/Warningnaming bug originates.Draft for review — the open naming question (
WARNING/NOTICE/HINTvs alternatives) is still being settled with @magnus-schjolberg.Bump
Related