Hey Rodney,

Well, the (technical) impact is not that big, as you could configure the plugin to not use the controller. But regarding security scaffolding, it's heavily based on the controller in combination with aerogear security, the client security scaffolding will remain unchanged (but there wasn't happen  very much things there for the moment).

So, basically if we want to offer the same (backend) security scaffolding features but without the controller, the refactoring could be bigger and could evolve changes on other third party forge plugins (like the REST plugin).

Seb




On Mon, Jun 10, 2013 at 9:37 PM, Rodney Russ <rruss@redhat.com> wrote:
Sebastien,

Does this affect the Aerogear scaffolding plugin in Forge?

-Rodney


To be honest, I'm not sure it's relevant to make the distinction between Controller with or without REST support. We are in 2013, I'm not using the Controller to render JSPs, especially not in a mobile scope  with our Front End libraries which are REST focused. I was using the controller because  I could describe with a fluent API my (rest) routes and integrate seamlessly my security concerns.



On Mon, Jun 10, 2013 at 6:15 AM, Daniel Bevenius <daniel.bevenius@gmail.com> wrote:
 > I think a combination of this e-mail and parts of Dan's would have been nice before we got an e-mail stating that the deprecation was happening and asking how we should go about it.
I'm to blame for this not being as open as it probably could have. The line of what is public and what in internal was a little blurred, and I opted to be careful, so the discussions have been mostly over private channels.

>The concern I have with leaving the REST bits as they are is confusion in what should be used, and when.  If we're not going to be enhancing the REST support, we should deprecate it imo.
I agree with this. 




On 10 June 2013 05:07, Kris Borchers <kris@redhat.com> wrote:
Though I am mostly on board with the idea of deprecating the REST bits of controller, I guess what I'm trying to figure out is when this was decided. I heard some passing mentions of controller getting beyond its initial scope, etc. but a discussion around deprecation/removal never happened that I am aware of. I think a combination of this e-mail and parts of Dan's would have been nice before we got an e-mail stating that the deprecation was happening and asking how we should go about it.

On Jun 9, 2013, at 10:01 PM, Jay Balunas <jbalunas@redhat.com> wrote:

Some of my thoughts:

The concern I have with leaving the REST bits as they are is confusion in what should be used, and when.  If we're not going to be enhancing the REST support, we should deprecate it imo.

If the controller's MVC functionality is going to move forward (i.e. not be deprecated) we should plan for its future carefully so that it works well with the REST support provided by JAX-RS /RESTEasy.  I'm just not sure of the priority over other things atm.  So it might be the type of thing where we take it where we can, and see if there is community interest in moving it forward.

I certainly would like to see a solid MVC be standardized in EE(X), and it might be in the future.  However, there are a lot of different  mvc frameworks out there and I'm concerned over trying to create another, especially if it seems our functionality is tied to it.

Either way, we should update our examples to show how to use our technology with RESTEasy, with or without the controller and/or other mvc options. So that users know they have options here.


On Jun 9, 2013, at 11:40 AM, Kris Borchers wrote:

I'm not sure deprecation is necessary but I would vote with Bruno to just leave the REST bits as is and ensure nothing else depends on them. Then we just make a note in the README that we have discontinued development and support on them. Then remove any references from AeroGear.org and only talk about our other server side bits.

On Jun 9, 2013, at 9:03, Bruno Oliveira <bruno@abstractj.org> wrote:

Ok, I almost had a heart attack here :) I'd say leave the REST bits as is and give to people a choice.

-
abstractj

On Jun 9, 2013, 10:57 AM, Daniel Bevenius wrote:

>Deprecate?! Why?! Do we have a lightweight MVC to replace?
I should have written, deprecate the RESTful part of AeroGear controller, as this is really what has been discussed.
If it makes sense to keep the MVC parts we can certainly do that. 




On 9 June 2013 15:51, Bruno Oliveira <bruno@abstractj.org> wrote:
Aloha Daniel,


So how do we make it visible that we have deprecated AeroGear-Controller?

Deprecate?! Why?! Do we have a lightweight MVC to replace?

-
abstractj

On Jun 9, 2013, 8:25 AM, Daniel Bevenius wrote:

Originally, AeroGear Controller was an MVC only implementation, where requests could be forwarded to different types of views. Later we added support for RESTful endpoints which was very basic to start with, but more and more feature requests have come which is the reason for creating this email.

We did not set out to implement the RESTful support our selves, instead the goal was to use RestEasy but as we required to be able to programmatically add endpoints which was not an option at the time (but support for this does now exist in RestEasy).

AeroGear controller's REST support started out very simple but as time passed more request for things that are taken for granted in a JAX-RS implementation were being asked for. We were moving toward something equivalent to a JAX-RS implementation with regards to what we supported. Adding more of these features added to the complexity of the Controller as we needed to figure out good ways to provide the features requested, and this also brings up the question if we should be doing this. The controller is not specific to mobil, as any backend can be used with the client SDKs. Also, there are plenty of options to implement RESTful backends, in Java using RestEasy for example, or in a different language. 

And after some discussions it has been decided that we should focus are efforts in other places when it comes to our server side offerings, for example the Unified Push Server and SimplePush Server etc. 

So how do we make it visible that we have deprecated AeroGear-Controller? 

A clear notice in the README.md is a good start. Should we mark every class as deprecated as well?

These are just suggestions and if you have other ideas please let us know. 

/Dan


_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev

_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev
_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev

_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev


_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev


_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev


_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev


_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev