-
Notifications
You must be signed in to change notification settings - Fork 2.7k
Migration Guide 1.3
You need to upgrade Maven to 3.6.2+ due to a bug in Maven resolver.
You need JUnit 5.6+. It's included in our BOM but if you specify the version manually, you need to upgrade.
REST Assured was updated to 4.2.
If you are tuning the Jackson ObjectMapper
, you need to change:
RestAssured.objectMapper(objectMapper);
to:
RestAssured.config().getObjectMapperConfig().defaultObjectMapper(objectMapper);
as they removed the shortcut.
The class loading model for dev and test mode has changed, which should fix a number of issues around class loading that have been reported. Most applications should continue to work as normal with no changes required, however there are some changes that may effect testing. The most visible change is that @BeforeAll
methods are now always run after Quarkus has started, so they can't be used to start resources Quarkus needs before boot. If you need to start resources before Quarkus has booted you can use the @QuarkusTestResource
annotation to specify a QuarkusTestResourceLifecycleManager
that can start and stop the resources.
19.3.1 and 20.0.0 are supported. Older versions are not.
Both JDK 8 and JDK 11 should work but keep in mind that GraalVM JDK 11 is a tech preview.
If you use Hibernate Validator in a multi-module project with constrained beans being in a library module, you need to be sure this library module is included in the bean archive index by adding a src/main/resources/META-INF/beans.xml
empty file.
Otherwise your constraints will be ignored.
Bytecode-enhanced bi-directional association management is no longer activated. It was not comprehensive and we decided to embrace the principle of least surprise and disable it. Make sure to update both sides of your relationships explicitly in your code.
Some info about Bytecode-enhanced bi-directional association management
The Emitter
interface has slightly changed. First, the interface has moved to org.eclipse.microprofile.reactive.messaging.Emitter
, as the interface is now part of the Reactive Messaging specification.
Also, sending Message
has changed. The Emitter generic parameter is the payload type and cannot be the Message
type:
// Before
@Inject @Channel("channel") Emitter<Message<String>> emitter;
// Now
@Inject @Channel("channel") Emitter<String> emitter;
emitter.send("foo");
emitter.send(Message.of("foo"));
Sending a payload returns a CompletionStage
completed when the message is acknowledged.
The KafkaMessage
class has been renamed into KafkaRecord
to be closer to the Kafka lingo.
The @Stream
annotation has been replaced by the @Channel
annotation.
In 1.3, a new reactive programming API has been introduced. This API named Mutiny (https://smallrye.io/smallrye-mutiny/) replaces the Axle and Reactive Streams Operators models (Reactive Streams and Completion Stage). The previous models are still functional, but deprecated and will be removed in the future. Refer to the various guides to show the differences and use the new API.
A few methods in the Search DSL have been renamed:
-
searchSession.search(...).predicate(...)
becomessearchSession.search(...).where(...)
. -
searchSession.search(...).asProjection(...)
becomessearchSession.search(...).select(...)
. -
searchSession.search(...).asProjections(...)
becomessearchSession.search(...).select(...)
. -
searchSession.search(...).asEntity(...)
becomessearchSession.search(...).selectEntity()
. -
searchSession.search(...).asEntityReference(...)
becomessearchSession.search(...).selectEntityReference()
.
The methods with the old name are still present, though deprecated. They will be removed before Hibernate Search 6.0.0.Final is released.
Elasticsearch indexes are now accessed through aliases only.
Hibernate Search relies on two aliases for each index: one for writes (<myindex>-write
) and one for reads (<myindex>-read
).
As a result, you will need to either create the two aliases, or drop your indexes and have Hibernate Search re-create empty indexes on startup: it will create the aliases.
See this section of the reference documentation for more information about index layout.
The names of synchronization strategies for automatic indexing have changed:
-
queued
=>async
. -
committed
(the default) =>write-sync
(still the default). -
searchable
(commonly used for tests) =>sync
(still recommended for tests only).
So if your configuration looked like this:
quarkus.hibernate-search.automatic_indexing.synchronization.strategy = queued
You should now have this:
quarkus.hibernate-search.automatic_indexing.synchronization.strategy = async
And if it looked like this:
quarkus.hibernate-search.automatic_indexing.synchronization.strategy = searchable
You should now have this:
quarkus.hibernate-search.automatic_indexing.synchronization.strategy = sync
See this section of the reference documentation for more information about synchronization strategies.
In 1.3, the configuration properties of the Kubernetes extension have changed in the following ways:
- They are now prefixed with
quarkus.
- Nested property arrays have been converted to maps.
For example the old property kubernetes.replicas
is now
quarkus.kubernetes.replicas
. This applies to all properties starting with
kubernetes
, openshift
and knative
. For compatibility reasons the old properties will still work but the new ones will always take precedence.
The other change affects properties that contained nested arrays,
like env-vars
and labels
:
For example, labels used to be configured like:
kubernetes.labels[0].name=foo
kubernetes.labels[0].value=bar
But the new properties are configured like:
quarkus.kubernetes.labels.foo=bar
The same applies to complex objects too:
kubernetes.env-vars[0].name=FOO_
kubernetes.env-vars[0].value=BAR
kubernetes.env-vars[0].name=FOO2
kubernetes.env-vars[0].secret=bar-secret
kubernetes.env-vars[0].name=FOO3
kubernetes.env-vars[0].config-map=bar-cfg
They can now be configured:
quarkus.kubernetes.env-vars.foo=value
quarkus.kubernetes.env-vars.foo2.secret=bar-secret
quarkus.kubernetes.env-vars.foo3.config-map=bar-cfg
The following unused and long-time deprecated classes were deleted from the codebase:
-
io.quarkus.test.junit.SubstrateTest
(Replaced withio.quarkus.test.junit.NativeImageTest
) -
io.quarkus.test.junit.DisabledOnSubstrate
(Replaced withio.quarkus.test.junit.DisabledOnNativeImage
) -
io.quarkus.test.junit.DisabledOnSubstrateCondition
(Replaced withio.quarkus.test.junit.DisabledOnNativeImageCondition
) io.quarkus.infinispan.client.runtime.Remote
- All classes in package
io.quarkus.deployment.builditem.substrate
(Replaced withio.quarkus.deployment.builditem.nativeimage
)
More information: https://github.com/quarkusio/quarkus/pull/5596