I prefer the idea that there's a class that manages one instance and that's all 
it does, and another object (or objects) to handle collections of those 
objects. I think separating them out makes for better reusability. If we'd like 
to have a save method I'm okay with that, but don't really see the need. 

Sent from my iPhone

On Sep 27, 2011, at 7:32, José Rodolfo Freitas <[email protected]> 
wrote:

> Sorry for not being able to post my gist. After writting some "sketches", I 
> realized that I don't have it very clear in my mind yet.
> 
> just to move from zero I posted something:
> 
> https://gist.github.com/1245041
> 
> It's a really simple example on how we could avoid inheritance,  but it's not 
> contemplating a lot of needed features,
> so I'm not even near to be convinced with that gist.
> 
> btw, the @AutoHome approach seems very nice.
> 
> 
> On Thu, Sep 22, 2011 at 2:12 PM, Dan Allen <[email protected]> wrote:
> Exactly. flush() has a specific purpose and really doesn't belong in 
> boilerplate code.
> 
> - Dan Allen
> 
> Sent from my Android-powered phone:
> An open platform for carriers, consumers and developers
> 
> On Sep 22, 2011 11:19 AM, "Max Rydahl Andersen" <[email protected]> 
> wrote:
> > just one comment:
> > 
> > Calling .flush() on every alteration really should not be promoted as a 
> > good practice.
> > 
> > /max
> > 
> > On Sep 22, 2011, at 24:22, Dan Allen wrote:
> > 
> >> Here's some additional feedback I received from a community member a while 
> >> back...to merge it into this thread.
> >> 
> >> (begin feedback)
> >> 
> >> ...from being burned from 3 seam based customers with apps and 
> >> maintenance. The "Home" or any other name should be just be put into a 
> >> grave and slowly cast away to sea ;). It is too heavy and complicated and 
> >> just about anything inherited (extends) truly causes heartache[Favor 
> >> Composition over inheritance: Effective Java]. The current seam home has a 
> >> few super classes above the home and when you try to unit test it (the 
> >> standard definition of unit-testing including isolation) you get the "No 
> >> Active Application Context Found (if I remember it right). That happens 
> >> because it is tightly coupled with the application. But not to be hard on 
> >> Home, I do realize the history of the home object and know it was 
> >> developed when EL had no parameters. So I have learned a lot since then 
> >> and I here are some things that I can impart to Seam 3.  
> >> 
> >> 1. My "Home" now is a "ServiceBean", and I have one for each "Major" 
> >> entity, see below. I have really stewed over this over months and months, 
> >> and the "Home" of "ServiceBean" should be kept small, focused, reusable, 
> >> tested and untouched. It's only task is to update, persist, possibly 
> >> remove, or some other functions that are required. In my example below I 
> >> have custom close action. Notice also that although these beans are 
> >> stateful that doesn't mean everything should be, so in these methods I 
> >> have the parameter of what is being needed to be updated, and not a field. 
> >> In other words I don't have @In private Job job, I opted for public 
> >> boolean update(job). Mostly because, again, I want to make this service 
> >> bean reusable so whether I have a #{newJob}, #{copyOfAJob}, or 
> >> #{managedJob} or whatever component of job I need to work on I only need 
> >> one jobServiceBean to cater to all my jobs, in whatever conversation I am 
> >> using. I also fire events from here if I need to do that. !
> > After this is tested, and what I need I usually don't touch it anymore. If 
> > I need to enhance I either use a decorator pattern around it, or enhance it 
> > in an @Observer. I'll email about that later.
> >> 
> >> @Name("jobServiceBean")
> >> @Scope(ScopeType.CONVERSATION)
> >> public class JobServiceBean implements JobService {
> >> private EntityManager entityManager;
> >> private StatusMessages statusMessages;
> >> 
> >> @In
> >> public void setEntityManager(EntityManager entityManager) {
> >> this.entityManager = entityManager;
> >> }
> >> 
> >> @In
> >> public void setStatusMessages(StatusMessages statusMessages) {
> >> this.statusMessages = statusMessages;
> >> }
> >> 
> >> public boolean update(Job job) {
> >> this.entityManager.flush();
> >> this.statusMessages.add(StatusMessage.Severity.INFO, "Successfully updated 
> >> job {0}", job.getName());
> >> return true;
> >> }
> >> 
> >> public boolean close(Job job) {
> >> job.setJobStatus(JobStatus.CLOSED);
> >> this.entityManager.flush();
> >> this.statusMessages.add(StatusMessage.Severity.INFO, "Successfully closed 
> >> job {0}", job.getName());
> >> return true;
> >> }
> >> }
> >> 
> >> 2. One thing you may have noticed from above that there is no 'instance' 
> >> field with corresponding getters or setters like the old 'Home'. So the 
> >> ServiceBean in my case is not a full crud, but CUD + your own business 
> >> methods. That's because that too should be decoupled because we never know 
> >> the source of the object is. Is the object created from a factory? from a 
> >> copy? is it a mapped component, a managed component? Creation of objects 
> >> or loading of objects, or the manufacturing of objects from factories 
> >> should be separate from the "home" or in my case the "ServiceBean".
> >> 
> >> (end feedback)
> >> 
> >> -- 
> >> Dan Allen
> >> Principal Software Engineer, Red Hat | Author of Seam in Action
> >> Registered Linux User #231597
> >> 
> >> http://www.google.com/profiles/dan.j.allen#about
> >> http://mojavelinux.com
> >> http://mojavelinux.com/seaminaction
> >> 
> >> _______________________________________________
> >> forge-dev mailing list
> >> [email protected]
> >> https://lists.jboss.org/mailman/listinfo/forge-dev
> > 
> > /max
> > http://about.me/maxandersen
> > 
> > 
> > 
> > 
> > _______________________________________________
> > forge-dev mailing list
> > [email protected]
> > https://lists.jboss.org/mailman/listinfo/forge-dev
> 
> _______________________________________________
> seam-dev mailing list
> [email protected]
> https://lists.jboss.org/mailman/listinfo/seam-dev
> 
> 
> _______________________________________________
> seam-dev mailing list
> [email protected]
> https://lists.jboss.org/mailman/listinfo/seam-dev
_______________________________________________
seam-dev mailing list
[email protected]
https://lists.jboss.org/mailman/listinfo/seam-dev

Reply via email to