drborges / appx-old Goto Github PK

This feature can also be used as a throttling mechanism for queries.

The following example shows a query that will hit datastore only once every 5 seconds.

runner := appx.NewCachedDatastore(context).
    Query(query).
    CachedAs(cacheID).
    ExpiresIn(5 * time.Seconds)

In case cache id is not specified the query will not be cached at all.

Counting matched entities:

totalMatches, err := runner.Count()

Getting the matched results:

err := runner.Results(&results)

Getting the first result:

err := runner.Result(&result)

Note Reusing the same cached runner to perform different operations such as Count, Results, Result will cause the cached data to be overwritten.

See this post for more information

Perhaps need to change the API for consistency sake and have for appx.Cacheable something similar to appx.Persistable#KeyMetadata:

type Cacheable interface {
  CacheMetadata() *CacheMetadata
}

type CacheMetadata struct {
  ID string
  ExpiresAt time.Duration
}

This would greatly simplify the way caching entities is handled by appx.CachedDatastore, avoiding the need of using the nasty appx.CacheableEntity workaround

LoadAll needs to:

• Load entities in batch from the cache
• Fall back to appx.CachedDatastore#LoadAll in case none of the entities are cached
• Fall back to datastore for each ErrCacheMiss
• Support appx.CacheMissQueryable entities
• Support loading over 1000 entities at once?

In case an entity is not present neither in the cache nor in datastore, an error should be returned.

For insights check this out

Currently it is used the JSON codec by default. Need to provide a way of users to switch codecs as they see fit.