#[cache_shared]Expand description
Caches a server function whose answer is the same for every visitor: at the CDN, on the server, or both.
Place it above the route attribute:
ⓘ
#[g3_cache::cache_shared(cdn = 300, server = "5m")]
#[get("/api/trending?media_type", db: Db)]
pub async fn get_trending(media_type: Option<MediaType>) -> Result<Vec<Media>> { .. }§Arguments
cdn = <seconds>: CDN caching. Successful responses are markedpublic, max-age=60, s-maxage=<seconds>(seeg3_cache::cdn).server = "<duration>": server caching, e.g."30s","5m","1h","1d". The whole function’sOkanswer is kept per set of arguments; concurrent misses share one call and errors are not kept.capacity = <entries>: the most answers the server cache keeps. Default 10 000.trust_extractors: skips the extractor check below, for an extractor whose name looks per-visitor but isn’t.
At least one of cdn and server is required.
§Checked at compile time
- The function uses
#[get]: a CDN never caches other methods, and shared data is read, not written. - No extractor in the route attribute is named like per-visitor state
(
session,auth,user,cookie,token,claims,header…), and the body doesn’t callFullstackContext. Either would make the answer depend on who asks, and sharing it would show one visitor’s data to everyone. - The function is
asyncand returns aResult.
§Not checked
A function can depend on the visitor in ways no macro can see: a query that uses the database session’s user, a global, a value from an extractor with a neutral name. Only share functions whose answer comes entirely from their arguments and public data.