We now support map columns that use mapped user-defined types and converted, non-primitive types in their keys and values. Map columns can be used for schema generation, column and key/value types are derived from the declared types by inspecting whether they are either UDTs or they can be converted by a custom registered converter.
class Supplier {
@Id String id;
Map<Manufacturer, List<Currency>> currencies;
}
@UserDefinedType
class Manufacturer {
String name;
}
class UDTToCurrencyConverter implements Converter<UDTValue, Currency> {
// …
}
class CurrencyToUDTConverter implements Converter<Currency, UDTValue> {
// …
}
We now support TupleValue as simple Cassandra type within domain objects along with table schema generation. Tuple columns must be annotated with @CassandraType to derive the according Cassandra column type for schema generation.
@Table
class Person {
@Id
String id;
@CassandraType(type = Name.TUPLE, typeArguments = { Name.VARCHAR, Name.BIGINT })
TupleValue tupleValue;
}
We now provide an alternative API for CassandraOperations that allows defining operations in a fluent way. FluentCassandraOperations reduces the number of methods and strips down the interface to a minimum while offering a more readable API.
// select with filter query and projecting return type
template.query(Person.class)
.as(Jedi.class)
.matching(query(where("firstname").is("luke")))
.all();
// insert
template.insert(Person.class)
.inTable(STAR_WARS)
.one(luke);
// update
template.update(Person.class)
.apply(update("firstname", "Han"))
.matching(query(where("id").is("han-solo")))
.all();
// remove all matching
template.delete(Jedi.class)
.inTable(STAR_WARS)
.matching(query(where("name").is("luke")))
.all();
We now support exists and count projections derived from query methods and on Template API level.
Exists queries fetch rows limiting to a single row to optimize fetching.
Exists queries also support count results, i.e. a single row with a single numeric column is considered a count result. It's evaluated by comparing the number being greater to zero.
boolean exists = template.exists(Query.query(where("firstname").is("Walter")), Person.class);
long count = template.count(Query.query(where("lastname").is("White")), Person.class);
interface PersonRepository extends Repository<Person, String> {
long countByLastname(String lastname); // derived query
boolean existsByLastname(String lastname); // derived query
@CountQuery("SELECT COUNT(*) from users WHERE lastname = ?0")
long countQueryByLastname(String lastname); // String-based query
@ExistsQuery("SELECT * from users WHERE lastname = ?0 LIMIT 1")
boolean existsQueryByLastname(String lastname); // String-based query
}
Remove superfluous throws declarations.
We now no longer use a Scheduler to offload ResultSet's blocking paging but request the next result page asynchronously by adapting ResultSet.fetchMoreResults(). Result pages are requested once all elements of the previous ResultSet are emitted and the row publisher completes successfully. The Scheduler is no longer required.
The request progress is stored in a MonoProcessor to extend the result stream. Increase visibility of utility methods to avoid synthetic accessor creation.
We now resolve user-defined type references to type stubs when constructing the create specification for a user-defined type instead of looking up the type from Cassandra if the particular property is annotated with @CassandraType(type = UDT). This applies to user-defined types nested in the to-be-created type. Stubbing is necessary to not prevent user-type creation during create specification construction. The actual execution happens after all creation specifications are built.
We now support keyspace alteration using XML configuration during after CassandraClusterFactoryBean initialization.
Previously, ALTER keyspace actions resulted in IllegalStateException.
