Search code examples
javagoogle-app-enginespring-roodatanucleus

Roo 1.1.5 super basic application is buggy

I'm using Roo 1.1.5, with Eclipse STS, here is what I do: I create a new Spring Roo Project from Eclipse, and in the Roo Shell, I type the following:

persistence setup --provider DATANUCLEUS --database H2_IN_MEMORY 
entity --class ~.domain.Task
field string --fieldName description --notNull --sizeMin 3 --sizeMax 512
field boolean --fieldName completed --notNull
controller all --package ~.controller

A more verbose description is in my blog, but what reported above is basically what I do.

In a web browser I can create and list tasks just fine, but if I try to edit (update) a Task, instead a new duplicated task is created... Possible that a bug like this has slipped in? Or is it something I'm doing wrong?

UPDATE: to be faster I've run the script from the command line and tried with all persistence providers.

$ roo
roo> project --topLevelPackage task --java 6 --projectName Task
roo> persistence setup --provider DATANUCLEUS --database H2_IN_MEMORY 
roo> entity --class ~.domain.Task
roo> field string --fieldName description --notNull --sizeMin 3 --sizeMax 512
roo> field boolean --fieldName completed --notNull
roo> controller all --package ~.controller
roo> exit
$ mvn tomcat:run #or mvn jetty:run

Here are the results:

DATANUCLEUS AND DATANUCLEUS_2 WITH H2

Roo + Datanucleus + H2 results in a buggy application, updating results in duplicated records.

ECLIPSELINK WITH H2

Build failure

The following artifacts could not be resolved: org.eclipse.persistence:eclipselink:jar:2.2.0, org.eclipse.persistence:javax.persistence:jar:2.0.3: Could not find artifact org.eclipse.persistence:eclipselink:jar:2.2.0 in spring-maven-release (http://maven.springframework.org/release)

OPENJPA WITH H2

Works, with warnings. Example:

WARN [main] openjpa.Runtime - An error occurred while registering a ClassTransformer with PersistenceUnitInfo: name 'persistenceUnit', root URL [file:/home/stefano/workspace/task4/target/classes/]. The error has been consumed. To see it, set your openjpa.Runtime log level to TRACE. Load-time class transformation will not be available.

HIBERNATE WITH H2

works

DATANUCLEUS WITH GOOGLE APP ENGINE

Works...

project --topLevelPackage task --java 6 --projectName Task
persistence setup --provider DATANUCLEUS --database GOOGLE_APP_ENGINE \
    --applicationId appid
entity --class ~.domain.Task
field string --fieldName description --notNull --sizeMin 3 --sizeMax 512
field boolean --fieldName completed --notNull
controller all --package ~.controller
exit
mvn gae:run
Solution

  • Are you perhaps hitting this JIRA ROO-1467. The JPA spec leaves many things open to interpretation. In DataNucleus 3 there is a flag to get various different types of behaviour for merge to cater for all eventualities.