Context
This issue tracks a potential improvement in how reasoning content is extracted from chat messages.
Background
In PR #482, a suggestion was made to improve the reasoning extraction by adding a fallback mechanism or using the centralized helper function:
- Current implementation only extracts reasoning from
finalMessage.parts without fallback
- A suggestion was made to either use the
extractReasoningContent helper or reinstate the fallback
Investigation Needed
Before implementing the suggested change, we need to:
- Verify if the UI correctly handles
undefined/null thinking parts
- Determine if the current implementation might lose data in edge cases
- Evaluate the impact of centralizing reasoning extraction logic
References
CC @AndlerRL