This is an automated email from the ASF dual-hosted git repository. rombert pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/sling-whiteboard.git
commit 3232af6e2c709b6eb9fcbb99d9268c4d90236147 Author: Robert Munteanu <[email protected]> AuthorDate: Wed Apr 22 15:19:06 2026 +0200 fix(skill_evals): use overall scorer This makes scoring, otherwise the 'parent_version_correct' scorer was unused and that one did not reflect the final result. --- skill-evals/src/skill_evals/jcr_js_nodetypes/scorer.py | 17 +++++------------ 1 file changed, 5 insertions(+), 12 deletions(-) diff --git a/skill-evals/src/skill_evals/jcr_js_nodetypes/scorer.py b/skill-evals/src/skill_evals/jcr_js_nodetypes/scorer.py index d6cc23ee..1989e1bc 100644 --- a/skill-evals/src/skill_evals/jcr_js_nodetypes/scorer.py +++ b/skill-evals/src/skill_evals/jcr_js_nodetypes/scorer.py @@ -10,13 +10,7 @@ POM_PATH = f"{REPO_DIR}/pom.xml" PARENT_VERSION_RE = re.compile(r"<parent>.*?<version>([^<]+)</version>", re.DOTALL) -@scorer( - metrics={ - "parent_version_correct": [accuracy(), stderr()], - "build_passed": [accuracy(), stderr()], - "overall": [accuracy(), stderr()], - } -) +@scorer(metrics=[accuracy(), stderr()]) def parent_pom_update() -> Scorer: async def score(state: TaskState, target) -> Score: expected_version = None @@ -52,17 +46,16 @@ def parent_pom_update() -> Scorer: explanation += f"mvn stderr (tail):\n{stderr_tail}\n" return Score( - value={ - "parent_version_correct": parent_version_correct, - "build_passed": build_passed, - "overall": overall, - }, + value=overall, answer=f"parent_version={actual_version}, build_success={build_result.success}", explanation=explanation, metadata={ "expected_parent_version": expected_version, "actual_parent_version": actual_version, "mvn_returncode": build_result.returncode, + "parent_version_correct": parent_version_correct, + "build_passed": build_passed, + "overall": overall, }, )
