Skip to main content

request_grouping

semantic_diff::grouper::llm

Function request_grouping 

Source 
pub async fn request_grouping(
    backend: LlmBackend,
    model: &str,
    hunk_summaries: &str,
) -> Result<Vec<SemanticGroup>>
Expand description

Invoke the LLM backend to group hunks by semantic intent.

Prompts are piped via stdin to prevent process table exposure of code diffs. Uses tokio::process::Command::spawn() so that aborting the JoinHandle drops the Child, which sends SIGKILL (critical for ROB-05 cancellation).