By Christian Beikov on 27 December 2021
We are happy to announce the fourth bug fix release of the 1.6 series. This release again contains not only bug fixes, but also some new features and bug fixes that I would like to highlight.
Jakarta artifacts
As of this release, we are publishing artifact variants with an artifact id suffix -jakarta which, as the name suggests, are Jakarta EE compatible.
This means that these artifacts depend on and use the jakarta.* namespace. We will most likely continue to ship separate artifacts for Jakarta EE compatibility until version 2.0.
With version 2.0 we will switch our main artifacts to use the jakarta namespace and instead produce -javaee artifacts for some time.
Ordered Set-Aggregate functions
Blaze-Persistence now supports ordered set-aggregate functions and the JPQL.Next language was extended to support the WITHIN GROUP clause syntax.
We added support for the PERCENTILE_CONT, PERCENTILE_DISC and MODE ordered set-aggregate functions and most importantly,
we also added a function for the SQL standard LISTAGG ordered set-aggregate function as alternative to GROUP_CONCAT
which we emulate on all major databases that have a supported vendor specific variant of the string aggregation, just like we did for GROUP_CONCAT before.
Have fun with string aggregations like LISTAGG(e.name, ', ') WITHIN GROUP (ORDER BY e.name)
JPA Criteria extensions
A great addition was made in the JPA Criteria module which now supports defining a WINDOW, analogously to the SQL WINDOW clause.
A WINDOW can be applied to window functions, as well as aggregate functions which are then used as window functions, and ordered set-aggregate functions.
Another nice addition is the possibility to specify a FILTER clause predicate for aggregate and ordered set-aggregate functions.
Here is a quick example that aggregates cat names that have an age greater than 10 as comma separated list, grouped by owner.
BlazeCriteriaBuilder cb = BlazeCriteria.get(criteriaBuilderFactory);
BlazeCriteriaQuery<Tuple> tupleQuery = cb.createTupleQuery();
BlazeRoot<Cat> cat = tupleQuery.from(Cat.class, "c");
tupleQuery.multiselect(
cb.listagg(cat.get(Cat_.name), cb.literal(", "))
.filter(cb.greaterThan(cat.get(Cat_.age), 10L))
.withinGroup(cb.asc(cat.get(Cat_.name)))
);
tupleQuery.groupBy(cat.get(Cat_.owner));
Tuple actual = tupleQuery.createCriteriaBuilder(em).getResultList().get(0);New GraphQL integrations
While helping a user to get started with SPQR, a GraphQL framework that is popular in the Spring ecosystem, a SPQR example application was developed.
Since the integration with SPQR was non-trivial, we introduced a new module that serves as integration between Blaze-Persistence and SPQR named blaze-persistence-integration-graphql-spqr
along with a new documentation section describing the setup.
The user wanted to go further and make use of updatable entity-views through GraphQL mutations, which we didn’t have experience with so far.
It was very interesting to get into this and we are pretty happy with the way the integration turned out.
TLDR, GraphQL mutations are now supported for all GraphQL integration modules except for Netflix DGS, which can only be supported until version 4.6, as the 4.6 release removed support for customizing the deserialization. Please help us convince the maintainers of DGS to consider adding an SPI by commenting and voting on the GitHub discussion
Take a look into the example applications to get a sense for how this can work out with your favorite framework!
Entity-View deserialization through JSONB
Since SmallRye GraphQL, the most popular implementation of MicroProfile GraphQL, relies on JSONB and we wanted to add support for GraphQL mutations for all GraphQL integration modules, we decided to develop an entity-view integration for JSONB as well. Unlike Jackson, JSONB unfortunately doesn’t provide support for registering deserializer object per type, but instead requires to register a deserializer class per type, which means we have to generate these deserializer classes before registration. The unfortunate consequence of this is, that this is currently a JVM only feature. We have plans to create an SPI that allows influencing code generation, which will solve this problem.
See the documentation for details about the setup of the JSONB integration.
While working on the JSONB integration, we noticed that our assumption about Jackson being the de-facto standard JSON framework for JAX-RS to be wrong, as JSONB the only supported JSON serialization framework on MicroProfile GraphQL. This lead to a small breaking change.
JAX-RS integration split
Due to the addition of the JSONB integration, which also required a dedicated JAX-RS integration variant, we decided to split the blaze-persistence-integration-jaxrs artifact
and move out the Jackson framework specific parts into a separate module blaze-persistence-integration-jaxrs-jackson.
We are sorry for the inconvenience, but when updating to 1.6.4 you now have to additionally add the Jackson module dependency, as the blaze-persistence-integration-jaxrs now only contains the API classes.
Quarkus improvements/fixes
Just like in the CDI integration, we now fire a EntityViewConfiguration event that allows customization of configuration, registration of custom basic types etc.
We were informed that JSON serialization for entity views didn’t work for native images which was now fixed. Thanks for the report!
Merry Christmas and a happy new year
We hope you enjoy this late Christmas present and wish you all a happy new year!
Take a look into the changelog for a full list of changes and improvements.
Enjoy the release and stay tuned for the next one!