Tonbo's read and write path needs a lot of context (version, storage manager and lru cache), such as storage manager, it might be used in the bottom of the read path but be initialized in the top-level (db instances). In the current, we just simply pass through them from top-level to bottom-level, it is ugly and expose implementations to those context agnostic layers. Therefore, use a unified context component to manage those would be better.
Tonbo's read and write path needs a lot of context (version, storage manager and lru cache), such as storage manager, it might be used in the bottom of the read path but be initialized in the top-level (db instances). In the current, we just simply pass through them from top-level to bottom-level, it is ugly and expose implementations to those context agnostic layers. Therefore, use a unified context component to manage those would be better.