Different databases are now supported by means of Maven Profiles and Spring Profiles. There is one Profile for each supported database. The default Profile is equivalent to hql. There is a Maven Profile all-dbs, which runs the integration tests against all databases.
The databases need to be started outside the build. The build assumes the default configuration as I found it after `brew install <database>`
For now we support the databases mysql, postgres and hsqldb.
In order to make the new databases work setting of properties and reading generated ids was improved to do some simple conversions. This might be considered a first step towards DATAJDBC-104.
The project root contains some basic scripts for starting and stopping databases, as well as running a build against all supported databases. Integration tests using a database now use Rules instead of JUnit runners. This gives more flexibility when adding fancy stuff to the Tests in the form of other Rules.
Related issue: DATAJDBC-104.
Original pull request: #5.
Events now use the id as source. The id is encapsulated in a value object to support null ids in BeforeInsert events. Entity references in events are now Optionals. Dropped redundant Event suffix from events. Added "@since 2.0" to classes. Replaced constructors and getters with Lombok annotations. Simplified pom.xml. Extracted interface JdbcPersistentEntity and JdbcPersistentProperty. Integration tests use the Spring Test framework. Fixed test for entities with primitive id type.
Original pull request: #5.
The repository publishes events before and after inserting, updating and deleting entities, as well as after instantiation of entities. JdbcEvent ist the common super class of all events and makes the id and the entity available (if possible). Added issue id comments to tests from previous issues.
Original pull request: #5.
New instances get saved with an insert statement. Existing instances get updated. Also added some test to find certain corner cases that I feared may cause problems:
- ID properties being not editable (no setter and final).
- ID properties being primitive.
- ID properties not being named "id"
and fixed the issues resulting from those.
Original pull request: #5.
No longer using batch inserts, which won't (easily) hold up for long anyway, since we need to decide between update and insert anyway. If one wants to support batch insert one also has to check if any autoid columns are present, because those seem not to work with batch inserts with many or at least some drivers.
Related ticket: SPR-1836.
Original pull request: #5.
Creation of the necessary sql statements is now dynamic and driven in a trivial way by the entity.
Known issues with the solution that need to get fixed later:
- SQL generating code is in the repository and should go somewhere else.
- Mapping logic is very trivial and should go in a separate place.
Original pull request: #5.
Created a maven project based on Spring Data JPA. Created JdbcRepositoryFactory, which actually creates Repositories. Wired the construction of meta data. Implemented a dummy version of saving an entity to demonstrate availability of meta data. Included a simple test for exercising the JdbcRepositoryFactory and the resulting JdbcRepository.
Related pull request: #5.
Jens Schauder