Brainstorming on introducing "Stacks" ------------------------------------- Key: FORGE-2316 URL: https://issues.jboss.org/browse/FORGE-2316 Project: Forge Issue Type: Sub-task Reporter: Antonio Goncalves Fix For: 3.x Future At the moment Forge generates code without having a real notion of a technical stack. For example, creating a JPA Entity for a Java EE 6 application could be different from a Java EE 7 application : * The {{persistence.xml}} version if different (2.0 / 2.1) * The {{properties}} are different (e.g. schema generation in 2.1) * The syntax is different ({{List<MyEntity> m = new ArrayList<MyEntity>}} while it could use diamond syntax for a Java EE 7 stack) The stack could be choosen when the project is created : * {{project-new --named myproj}} : let's the developper use Forge without any special stack * {{project-new --named myproj --stack JavaEE7}} : the generated code will follow Java EE 7 h3. Possible stacks A stack would bundle certain facets. Therefore we could have as many stacks as needed. As an example, we could have the following stacks : * Java EE 6 : JPA 2.0, CDI 1.0, JSF 2.0, Bean Validation 1.0, JTA 1.1 * Java EE 7 : JPA 2.1, CDI 1.1, JSF 2.2, Bean Validation 1.1, JTA 1.2, Batch 1.0 * Micro Java EE 7 Service: JPA 2.1, CDI 1.1, JSF 2.2, Bean Validation 1.1, JTA 1.2, Batch 1.0 (but with special Microservice configuration) * Java EE 8 : JPA 2.1, CDI 2.0, JSF 2.3, Bean Validation 1.2, JTA 1.2, Batch 1.0, JCache 1.1, MVC 1.0 * Spring 3.x * Spring 4.x * .... h3. Differences in having stacks h4. Filtering commands When you pick up a Stack, it gives you access or not to certain commands. For example, if you create a project with a Java EE 6 stack, you won't have access to any Batch, MVC... commands {code:title=Java EE 6 stack} $ project-new --named myproj --stack JavaEE6 $ jpa-new-entity --named MyEntity {code} {code:title=Java EE 7 stack} $ project-new --named myproj --stack JavaEE7 $ jpa-new-entity --named MyEntity $ mvc-new-controller --named MyController // only available on EE7 {code} h4. Setting up command versions When you pick up a Stack, it filters the commands that have the right version for the right stack. For example, if you create a Java EE 6 stack, you will get JPA 2.0 commands, for a Java EE 7 stack, the JPA 2.1 commands. Also, the {{version}} parameters will be disabled {code:title=No stack} $ project-new --named myproj $ jpa-new-entity --named MyEntity --jpaVersion 2.1 $ cdi-new-bean --named MyBean --cdiVersion 1.1 {code} {code:title=Java EE 6 stack} $ project-new --named myproj --stack JavaEE6 $ jpa-new-entity --named MyEntity // no --jpaVersion because it is automatically set to 2.0 {code} {code:title=Java EE 7 stack} $ project-new --named myproj --stack JavaEE7 $ jpa-new-entity --named MyEntity // no --jpaVersion because it is automatically set to 2.1 $ jpa-new-converter --named MyConvert // JPA Converters are only available in 2.1 {code} h4. API to define a Stack A stack could be defined with a fluent API such as : {code} stack("JavaEE7").includes(JavaEE7Facet).includes(JPA2.1Facet).excludes(SpringFacet) stack("JavaEE7").includes(JavaEE7Facet).includes(JPA2.1Facet).allowsFeature(WildflySwarm) stack("MyOwnStacks").includes(JavaEE7Stack).includes(MyFacet).includes(YourFacet) {code}