Skip to content
Orionis Framework

Cache

Cache results you can reconstruct: frequent queries, summaries or expensive calculations. Decide how long a value may become stale and when to invalidate it. The database keeps permanent information; caching avoids repeated work.

First operation from a controller

Section titled “First operation from a controller”

The default file store works with the normal Orionis installation. Set these application environment values before startup:

CACHE_STORE=file
CACHE_PREFIX=my-app

Its default directory is storage/framework/cache/data; the application process needs write access there. No Redis server or additional Python cache package is needed. Import the facade from orionis.support.facades. Save this controller in app/http/controllers/cache_example_controller.py:

from orionis.http import JSONResponse
from orionis.support.facades import Cache




class CacheExampleController:
    async def __call__(self) -> JSONResponse:
        key = "catalog:summary:v1"
        await Cache.set(key, {"products": 42}, ttl=300)
        return JSONResponse(await Cache.get(key))

Register its action in the configured route file:

from app.http.controllers.cache_example_controller import CacheExampleController
from orionis.support.facades import Route


Route.get("/cache-example", CacheExampleController)

The dictionary is cached for five minutes. Once expired or deleted, get() returns None. To calculate only on a miss, use remember() in the operations guide.

Service architecture

Section titled “Service architecture”
flowchart TD
  accTitle: Accessing the Orionis cache
  Config["config/cache.py"] --> Manager["CacheManager singleton"]
  Access["Cache or ICacheManager injection"] --> Manager
  Manager --> Repo["CacheRepository: applies the prefix"]
  Repo --> Backend["Selected driver"]
  Repo --> Locks["CacheLock"]
  Backend --> Values["Async reads and writes"]

The manager selects one of five native stores and retains its repository. The facade delegates to the same service resolved by the container. Value operations are asynchronous; the store() and lock() factories are synchronous. If your first facade operation selects a store, resolve cache = await Cache.resolve() first, then call cache.store("memory"). Its provider is deferred: importing Cache or booting the application does not guarantee resolution. Asynchronous calls such as await Cache.get(...) resolve it when awaited.

FileBasedCache is a separate synchronous API for artifacts derived from source files. It is not the file driver and does not use its TTL or prefix.

Continue with the guides

Section titled “Continue with the guides”