-
Notifications
You must be signed in to change notification settings - Fork 1.4k
Open
Labels
AI patchingFeature requests and bugs related to AI-based kernel bug fix generation.Feature requests and bugs related to AI-based kernel bug fix generation.enhancement
Description
LLM-generated text output (e.g. explanation as to why a race is benign) is not capped at any line lengths. By default LLM assumes the text will be reflown when rendered (which is reasonable).
As the result the text cannot be directly copy-pasted into kernel emails.
For patch descriptions I asked LLM to format to 72 cols. And it usually works, but I wonder if we should not bother LLM with such mechanical task (it can easily mess it as well I guess). Perhaps we should re-flow manually. We could do it once as part of workflow, or perhaps better to have a function for that in dashboard, and reflow where/as necessary.
Metadata
Metadata
Assignees
Labels
AI patchingFeature requests and bugs related to AI-based kernel bug fix generation.Feature requests and bugs related to AI-based kernel bug fix generation.enhancement