Skip to content

Separate L1 replication/hit/miss tracking from storage #63

@davidstrauss

Description

@davidstrauss

While writing SQLiteL1, it became apparent that APCu is still the best way to track aggregate hit/miss data and the event ID cursor. Unfortunately, this means duplicating code between APCuL1 and SQLiteL1. It would also be inappropriate to make this a general part of the L1 superclass, at least if the intention is supporting useful L1 implementation that lack APCu. It would also make testing difficult if L1 becomes coupled to APCu.

The best path forward is, probably, isolating lightweight counters and L1 event ID tracking into its own class so it's possible to use DatabaseL2 + SQLiteL1 + a "pool state" class (or similar).

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions