[seam-dev] seam-wicket -> wicket-weld

[Pete Muir]

On 17 Aug 2011, at 21:59, Ove Ranheim wrote:

> 
> On Aug 17, 2011, at 10:21 PM, Clint Popetz wrote:
> 
>> 
>> 
>> On Wed, Aug 17, 2011 at 3:17 PM, Ove Ranheim <oranheim at gmail.com> wrote:
>> Clint,
>> 
>> What is it specifically that you will benefit from taking this into Wicket? What plans do you have and what would be to the better?
>> 
>> Please enlighten me, but what's the problem for "the resource you're trying to convince" of not contributing under the Seam umbrella. Why is it a problem for the wicket team to stay in sync with releases? Isn't it approx a year between every Wicket release?
>> 
>> 
>> The idea is that if wicket 1.x is released, wicket-cdi 1.x will be released simultaneously.  In addition, as wicket evolves, the person most familiar with its evolution (Igor) will have responsibility for keeping the integration current.  Finally, wicket already supports other dependency injection frameworks (spring, juice) as wicket modules, so it makes sense for the cdi module to live alongside those, and will give cdi more exposure for those looking to use dependency injection in wicket.
>> 
> 
> There's a danger with two many "currents" and no well architected umbrella. I'd rather see many more frameworks being integrated on top of Solder. It's a concern that reusability will suffer due different strategies of implementation. What Solder, Faces, Servlet and more, really does well. Is to make a unified glue layer to fully integrate with different technologies.
> 
> I'm not too confident moving Wicket out in it's own Wicket CDI module. IMHO, it'll be an Igor vs a whole Seam Community empowerment where talking about here. Has Igor fully evaluated the Seam Solder eco-system? Will you make a new Wicket CDI Int, Servlet, Catch module too. Eventually, how well will a mixed platform play together in the future!? Is Wicket strategy to stay Wicket and no other than Wicket should get in.

I think you are making a lot of (negative) assumptions about the way this will work out. Clint is a long time contributor to Seam and Weld, and Igor has previously contributed to Weld. So I think we can have complete faith that the work they do will reuse what makes sense from Seam, and play nicely with the Seam ecosystem. I don't think Wicket is proposing to create an entire new set of CDI plugins, and I can't really read that into what Clint says.

There are a lot of benefits to a relevant technology owning the CDI integration. 

Firstly, expose. On the one hand, by placing it in Seam, we limit it's perceived applicability to "Seam" projects, and we don't encourage the wider Wicket community to try out CDI. OTOH if it lives in Wicket, we encourage that whole community to try out CDI, and perhaps take a look at Seam.

Secondly, release cycles. CDI is a backwards compatible, specified, API, whilst Wicket in this case is more flexible. Therefore taking CDI as the stable point makes a lot of sense. We tried the approach you suggest in Seam 2.x and it quickly becomes unmanageable and limits the size of the ecosystem. This was part of the motivation for developing CDI.

Thirdly, expertise. CDI is well documented and specified (not to say Wicket isn't, but if we generalize these arguments, this does often hold) and has a growing community of people who are expert at integration techniques, and to reaching out to third parties and helping them integrate with CDI. Having the Wicket team own the integration makes a lot of sense here too.

Finally, that there is actually nothing to stop the Wicket team making an official CDI integration without the Seam communities involvment at all. Much better to engage than build walls.

There are the disadvantages you mention, however I believe they are small, and can be mitigated almost entirely by Seam documenting third party integrations.

HTH understand the reasoning behind the Seam/JBoss philosophy here that it's better that the integration with CDI lives outside Seam where possible.

Pete