Struct fastly::cache::core::TransactionInsertBuilder

pub struct TransactionInsertBuilder { /* private fields */ }

A builder-style API for configuring a transactional cache insertion.

Implementations

impl TransactionInsertBuilder

pub fn vary_by<'a>( self, headers: impl IntoIterator<Item = &'a HeaderName> ) -> Self

Sets the list of headers that must match when looking up this cached item.

Note: These headers are narrowly useful for implementing cache lookups incorporating the semantics of the HTTP Vary header, but the APIs in this module are not suitable for HTTP caching out-of-the-box. Future SDK releases will contain an HTTP Cache API.

The headers act as additional factors in object selection, and the choice of which headers to factor in is determined during insertion, via e.g. InsertBuilder::vary_by. A lookup will succeed when there is at least one cached item that matches lookup’s cache key, and all of the lookup’s headers included in the cache items’ vary_by list match the corresponding headers in that cached item.

A typical example is a cached HTTP response, where the request had an Accept-Encoding header. In that case, the origin server may or may not decide on a given encoding, and whether that same response is suitable for a request with a different (or missing) Accept-Encoding header is determined by whether Accept-Encoding is listed in Vary header in the origin’s response.

pub fn initial_age(self, age: Duration) -> Self

Sets the initial age of the cached item, to be used in freshness calculations.

The initial age is Duration::ZERO by default.

pub fn stale_while_revalidate(self, duration: Duration) -> Self

Sets the stale-while-revalidate period for the cached item, which is the time for which the item can be safely used despite being considered stale.

Having a stale-while-revalidate period provides a signal that the cache should be updated (or its contents otherwise revalidated for freshness) asynchronously, while the stale cached item continues to be used, rather than blocking on updating the cached item. The methods Found::is_usable and Found::is_stale can be used to determine the current state of a found item.

The stale-while-revalidate period is Duration::ZERO by default.

pub fn surrogate_keys<'a>(self, keys: impl IntoIterator<Item = &'a str>) -> Self

Sets the surrogate keys that can be used for purging this cached item.

Surrogate key purges are the only means to purge specific items from the cache. At least one surrogate key must be set in order to remove an item without performing a purge-all, waiting for the item’s TTL to elapse, or overwriting the item with insert().

Surrogate keys must contain only printable ASCII characters (those between 0x21 and 0x7E, inclusive). Any invalid keys will be ignored.

See the Fastly surrogate keys guide for details.

pub fn known_length(self, length: u64) -> Self

Sets the size of the cached item, in bytes, when known prior to actually providing the bytes.

It is preferable to provide a length, if possible. Clients that begin streaming the item’s contents before it is completely provided will see the promised length which allows them to, for example, use content-length instead of transfer-encoding: chunked if the item is used as the body of a Request or Response.

pub fn user_metadata(self, user_metadata: Bytes) -> Self

Sets the user-defined metadata to associate with the cached item.

pub fn sensitive_data(self, is_sensitive_data: bool) -> Self

Enable or disable PCI/HIPAA-compliant non-volatile caching.

By default, this is false

See the Fastly PCI-Compliant Caching and Delivery documentation for details.

pub fn execute(self) -> Result<StreamingBody, CacheError>

Begin the insertion, returning a StreamingBody for providing the cached object itself.

For the insertion to complete successfully, the object must be written into the StreamingBody, and then StreamingBody::finish must be called. If the StreamingBody is dropped before calling StreamingBody::finish, the insertion is considered incomplete, and any concurrent lookups that may be reading from the object as it is streamed into the cache may encounter a streaming error.

pub fn execute_and_stream_back( self ) -> Result<(StreamingBody, Found), CacheError>

Begin the insertion, and provide a Found object that can be used to stream out of the newly-inserted object.

For the insertion to complete successfully, the object must be written into the StreamingBody, and then StreamingBody::finish must be called. If the StreamingBody is dropped before calling StreamingBody::finish, the insertion is considered incomplete, and any concurrent lookups that may be reading from the object as it is streamed into the cache may encounter a streaming error.

The returned Found object allows the client inserting a cache item to efficiently read back the contents of that item, avoiding the need to buffer contents for copying to multiple destinations. This pattern is commonly required when caching an item that also must be provided to, e.g., the client response.