Two design changes from current literature (research(2026-09))¶

Hunks now carry surrounding context. arXiv:2607.25851 (Rethinking Training Data for Generating Code Review Comments, read 2026-09-14) names three sources of misalignment between a code change and its review comment: semantic ambiguity, lack of actionability, and context dependence. The third is a real gap here, and it is the same root cause as a failure already measured: a bare hunk gave the model no way to know its indentation level, so a rewrite that was otherwise exactly right lost strict exact match. Context is prompt material only; the target stays the hunk and scoring compares only the hunk.

Corpus construction stays deterministic, deliberately. The field has moved toward LLM-based filtering for comment actionability, and that same paper reports up to a third of comments being vague or non-actionable. It also concludes that detecting misaligned training pairs remains challenging even with LLM-based approaches. An LLM filter would reintroduce exactly the reproducibility problem that kept LLM-as-judge out of the metric, so non-actionability is treated as a measured characteristic of the corpus to report, not a filter to apply. The deterministic filters stay: author comments, metadata pseudo-files, and hunk anchoring.