Patch release with the best_trial maximize fix (#258)? #259

Request

Could you publish a patch release (e.g. 0.1.1) that includes #258?

#253 (Study.best_trial ignores the study direction and returns the minimum) was fixed on main by #258 on 2026-09-17, but the latest release on PyPI is still 0.1.0 (2026-09-07), so pip install rustuna still ships the bug. A patch release was mentioned in #253; this is a gentle follow-up in case it helps with tracking.

Why it matters

With direction="maximize", best_trial silently returns the worst trial, so a rustuna study looks worse than the same Optuna study even though the search itself works correctly. I hit this in a real hyperparameter search (ROC-AUC, 20 trials): best_trial reported 0.750 while the actual best trial reached 0.829. It is easy to draw the wrong conclusion from that, especially with import rustuna as optuna.

Reproduction with rustuna==0.1.0 from PyPI

import rustuna

values = iter([0.80, 0.83, 0.75, 0.81, 0.78])

def objective(trial):
    trial.suggest_float("x", 0, 1)
    return next(values)

study = rustuna.create_study(direction="maximize")
study.optimize(objective, n_trials=5)

print(study.best_trial.number, study.best_trial.value)          # 2 0.75  <- the minimum
print(study.best_trials[0].number, study.best_trials[0].value)  # 1 0.83  <- correct

Workaround until the release

study.best_trials[0], max(study.trials, key=lambda t: t.value), or rustuna.converter.ToOptunaStudy(study).best_trial.

Environment

rustuna 0.1.0 (PyPI wheel cp311-abi3-macosx_11_0_arm64), Python 3.11 and 3.12, macOS on Apple Silicon.

Thank you for the project!