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
Purpose: Then we can afford to box the ops, and we can do a BoxedStore, which in the long term means that the Blobs protocol no longer needs a store type parameter. Also, we might be able to move the LocalPool into the file store.
The downside is that there is less code sharing between the mem and the file store, and writing a primitive mem store becomes more complex because we have to provide high level ops. But I think it is probably worth it just for getting rid of the type parameter.
The text was updated successfully, but these errors were encountered:
The current thinking is to get rid of the entire hierarchy and instead do a "protocol" that store implementers have to implement, consisting of the good old low level big enum of commaands with oneshot or mpsc senders for responses...
This will be slightly annoying for store implementers, especially if they don't want to support the full functionality. But it will be much nicer for store users. No more S type parameter everywhere, e.g.
Idea: make store ops more high level.
Purpose: Then we can afford to box the ops, and we can do a BoxedStore, which in the long term means that the Blobs protocol no longer needs a store type parameter. Also, we might be able to move the LocalPool into the file store.
The downside is that there is less code sharing between the mem and the file store, and writing a primitive mem store becomes more complex because we have to provide high level ops. But I think it is probably worth it just for getting rid of the type parameter.
The text was updated successfully, but these errors were encountered: