- Initial relese
- Deprecated
impl RolloutStrategy for f64
in favor of thePercent
rollout strategy
- Introduce the
run_result
method with special handling forResult
values - Remove
RolloutDecision::UseExperimental
- Remove
thesis_experiment_run_mismatch
metric in favor of the more generalthesis_experiment_outcome
- Upgrade to Tokio 1.0
- Upgrade to
metrics
>=0.22, <=0.24 - Add
thesis_experiment_run_duration
histogram for variant runtime - Allow
RolloutDecision
to be passed directly torollout_strategy