Niklas

The easier alternative is to move all API server tests to use PostgreSQL testcontainers instead of H2. Has the nice side effect that we can ban H2 from the project...

Raised https://github.com/datanucleus/datanucleus-rdbms/issues/494. Not really sure what areas of DataNucleus will need to be touched, so hoping the maintainer can nudge me in the right direction.

Turns out this can be solved with extensions: ```java @Persistent @Column(name = "DATA", jdbcType = "CLOB") @Extensions({ @Extension(vendorName = "datanucleus", key = "insert-function", value = "(?::JSONB)"), @Extension(vendorName = "datanucleus", key...

No implementation necessary for this one, see comment above.

Re-opening because we had to temporarily roll back this change.

#### Breaking Change: Some Notification Values Changed * Generally, `subject` objects passed to notification templates are Protobuf objects now, as defined in https://github.com/DependencyTrack/hyades/blob/main/proto/src/main/proto/org/dependencytrack/notification/v1/notification.proto * Shouldn't break anything by itself, but...

@VinodAnandan If we use the same technology for both API server and CLI, we have opportunities to share code among them. API models, validation, potentially CEL evaluation and more. If...

Implementation of this must also include a mechanism for coordinated cache invalidation.

A few example implementation of batching that I found: * https://github.com/apache/kafka/blob/3.6.0/connect/runtime/src/main/java/org/apache/kafka/connect/runtime/WorkerSinkTask.java * https://github.com/getsentry/batching-kafka-consumer/blob/master/batching_kafka_consumer/__init__.py * https://github.com/benthosdev/benthos/blob/main/internal/impl/kafka/input_kafka_franz.go

I just realized that Confluent's Parallel Consumer is doing exactly that: https://github.com/DependencyTrack/hyades/issues/346 Update: Confluent Parallel Consumer has no batch timeout behavior (doesn't wait for batches to become full). So ultimately...