Releases: graphql-java/java-dataloader
3.1.1
3.1.0
ValueCache is called in batch function
The 3.x version of DataLoader added a long wanted feature. The ability to have an external cache of values.
It was rightly made an async operation to get cache entries.
However we got the code design wrong and called the caching during the dataLoader.load() calls and this broke the implied contract between loading N keys and being in a position to dispatch the batch loader and retrive N keys in batch.
This release fixes that - it moves the async caching get to the batch function itself, which is truly async.
This had a side benefit as well. The ValueCache now has a getValues(List<K> keys) method that allows a batch worth of keys to checked at once. External systems sucha s REDIS have multiple GET APIs and hence the caching can be even more efficient
graphql-java will now be able to update to this version in order to use the modern data loader with external value caching.
3.0.1
3.0.0
Breaking Changes
- the
CacheMapsignature has been changed to correctly reflect that it is a cache ofCompletableFuture<V>and not just<V>
Other changes
-
a new
ValueCachelayer has been added to that allows values to be stored in external cache systems such as REDIS or long lived cross request systems like Caffeine -
A new last dispatched time is present on DataLoaders
-
An experimental
ScheduledDataLoaderRegistrythat allows new ways to dispatch dataloaders.
2.2.3
2.2.0
2.1.1
This adds new features into data loader to allow you to write more powerful batch loading functions
The first is returning a Map from batch loader function instead of an ordered list. This suits some use cases much more naturally
MappedBatchLoaderWithContext<Long, User> mapBatchLoader = new MappedBatchLoaderWithContext<Long, User>() {
@Override
public CompletionStage<Map<Long, User>> load(Set<Long> userIds, BatchLoaderEnvironment environment) {
SecurityCtx callCtx = environment.getContext();
return CompletableFuture.supplyAsync(() -> userManager.loadMapOfUsersByIds(callCtx, userIds));
}
};
DataLoader<Long, User> userLoader = DataLoader.newMappedDataLoader(mapBatchLoader);
// ...The second major change is that context can now be pushed into batch loading functions allowing you to get user credentials or database details for example
DataLoaderOptions options = DataLoaderOptions.newOptions()
.setBatchLoaderContextProvider(() -> SecurityCtx.getCallingUserCtx());
BatchLoaderWithContext<String, String> batchLoader = new BatchLoaderWithContext<String, String>() {
@Override
public CompletionStage<List<String>> load(List<String> keys, BatchLoaderEnvironment environment) {
SecurityCtx callCtx = environment.getContext();
return callDatabaseForResults(callCtx, keys);
}
};
DataLoader<String, String> loader = DataLoader.newDataLoader(batchLoader, options);