You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This release is backward compatible with the previous one.
Changed
Added ThreadMapX type that contains a Mutex for each value, allowing the methods Self::fold, Self::fold_values, and Self::probe to run more efficiently when there are concurrent calls to the per-thread methods (Self::with and Self::with_mut) by using fine-grained per-thread locking instead of acquiring an object-level write lock. On the other hand, the per-thread methods may run a bit slower as they require the acquision of the per-thread lock.