Develop Reference

Activator Play Framework with google app engine

I am wanting to develop a GAE application. However, I have used the play framework before and very much liked its layout. I was wondering whether the latest version (2.3 I think) works with Google App Engine? I have seen posts about this before. Some say it is not possible; I've seen a github project on porting the code over but no one has ever commented on the success of the project. I am concerned about losing efficiency/too much functionality.
Can it be done?
If not, any suggestions on frameworks (I know I am not really supposed to ask this second question on stackoverflow).

Have a look at this repo which enables you to package your Play app in a war that you can deploy onto GAE.
But you should consider the trade-offs before doing this. Play was designed to be stateless/Asynchronous and GAE runs on Java Servlet Spec. This means (generally):-
your Play app will not be truly asynchronous on GAE
you may reduce performance a bit because you are running in servlet container on GAE (depending on your app though you may not even be able to measure this so I wouldn't worry too much about it)
I have personally found that deploying my Play apps into Digital Ocean (I use Docker but you don;t have to) work really well for me. I have used GAE in the past but I wouldn't deploy a Play app there for the reasons stated above. Heroku is a nice alternative but gets a bit costly if you need to scale out.

I got Playframework to work on Goole App Engine (Flexible, Custom runtimes). It involves writing our own Docker image. The integration isn't seamless (e.g. default credential doesn't work, so we have to use a JSON credential).
See: http://tanin.nanakorn.com/blogs/345

Java web application development kit recommendation

I'm looking for good java MVC web application development kit (or framework), with servlet engine, database persistence layer and viewer layer included and ready to go? One example is TDK (Turbine Development Kit), but it's kinda outdated, and their M.E.T.A. is very hard to work with (it uses maven 1.0).
Thanks,
David

There's AppFuse and Roo if you're looking for "make it go".
If you're not dead-set on Java, there's Grails (and JRuby on Rails).

Give a shot to AppFuse. It will leverage either Struts 2 or Spring MVC for you, with a nice persistence layer.

There are many good 'stacks' that you could use, some of which have already been mentioned.
They all have their own strengths & weaknesses, and you really need to select a stack that best fits your requirements.
The Spring/Roo, AppFuse & Seam are all great, but they do come with a certain amount of... errm... bloat.
You may find that picking and mixing technologies actually provides you with a lighter weight and more performant solution, so long as the technologies you choose integrate relatively seamlessly.
If you're looking to package a full execution environment including the servlet engine as part of your application, Jetty is a good choice, and you can easily integrate it into a Maven build environment.
You may also want to investigate using HyperSQL for your database, which will allow you to prototype using an SQL compliant database without having to worry about installing a full database manager such as MySQL or PostgreSQL.

Alternative usage of Spring Roo

is Spring Roo supposed to be only a tool for rapid development of web applications, thanks to all that scaffolding, source code generation and similar stuff, such as grails for groovy.
Or can it be utilized as a source code generator / osgi platform for building enterprise applications ? What I mean is, that there are source code generators, that generate only domain model and light DAO layer based on metadata definition when starting a new project and you have to stick to the objective - create a web app. But there are also generators, that generate the entire DAO and service layer - fantastic for iterative dev process of bigger apps - and you are not restricted otherwise, you are just using it to build you enterprise app step by step. With this you can really build a huge enterprise applications / platforms that will be maintainable for a very long time.
Also the OSGI model could be employed in a way, that you might develop a core portal application for instance, with a package of social office addons that you may plug in, etc. etc.
My point is, that Java development needs to be "Rubyfied". I mean Ruby on Rails environment where you have everything unified and standardized. A platform that could rule the java world some time :-)

Since Thomas and #jhericks trashed Spring Roo I'm going to say I actually like it mainly because its not a framework. Roo also (now) support multi-maven modules.
To answer your question: Spring Roo is not framework. It is code generation.
So yeah it is a code generation platform not only for web apps but any Java app.
There are couple of reasons why some people don't like Roo:
It uses AspectJ
It uses Spring
It does Code Generation
After building several applications with Spring Roo I can say its greatest value is its ORM generating abilities and its DataOnDemand test generation.
Other than that Spring Roo is really just Spring *.
Frameworks like Play and Ruby on Rails are just not really the "Java Way". That is most Java (Web) Apps I see are combination of libraries and not a giant framework.

As far as I experienced, Spring Roo is meant to be a RAD/bootstrapping tool for web applications. There are some major features missing and not on the roadmap that would make it truly useful in an enterprise application context. An example being multi-module Maven support.
This rather polemic article and its follow ups summarize quite well how I feel about Roo and similar projects like Seam Forge.
If you are looking for something RoR-like for Java check out the Play Framework that currently seems to gain momentum.

How to develop Java webapps in a Ruby on Rails-like way?

We've got an app that needs to run in a Java app server. I've got to develop a webapp to go with it. I've started reading about Ruby on Rails, and I really like the approach: MVC, convention over configuration, tons of stuff taken care of for you.
I may be dreaming the impossible dream, but are there any Java webapp development frameworks out there that work in a similar way?
Straight JSP is miserable, because logic and presentation are mixed. Simple template systems, like Velocity and Freemarker don't do the database stuff. I don't know much about Spring.
Everybody here knows Java, and we don't have time before the next release cycle to learn an entirely new language and library, else I'd plunge into RoR in a serious way.

Not impossible at all. It's already been done.
Here are a few Java-equivalents:
Grails http://www.grails.org/
SpringRoo. http://www.springsource.org/roo
Spring MVC. http://www.springsource.org/
All are products of SpringSource. If you're getting serious about Java web development, I highly recommend you check them out.
Grails was actually inspired by Rails, hence the name, so you'll see many similarities in features and paradigms between the two, including MVC, convention-over-configuration, and code generation. So if you're looking to get right into Rails-like development in Java, I highly recommend Grails. I should note that Grails has more features than Rails, some of which I think Rails should adopt. An example is Grails provides a service layer to handle all business logic. You could easily implement a service layer in Rails, but its convenience we all love, and because Grails includes it, it is more convenient. I've hopped between Rails and Grails once upon a time, both great frameworks.
Keep in mind that you can do Rails development with Java as well, via JRuby, which is not a bad option either. You get all the features of the Ruby language (v. 1.8), along with the ability to integrate and use Java libraries within your Ruby code.
SpringRoo is SpringSource's new code generator framework, so you can use it with the Spring framework to aid in agile/rapid application development.
Grails is actually built on-top of the Spring framework and Groovy. Groovy is Java-derivative which looks more like a scripting language and provides all the meta/dynamic programming goodness that you see in languages such as Ruby and Python. However, Groovy allows you to write Java code as well, so there's no need to learn it, not entirely at least.

Have a good look at the the Play Framework. I think that it's the best way to develop Java web applications at the moment. Certainly the quickest. There is no deploy cycle. Save the code and refresh the browser.
It comes with Hibernate and a host of enhanced functionality that streamlines the development process. Totally REST based and stateless.
Add in jQuery and it feels like web heaven.

Is Grails a viable option for large scale enterprise web apps?

Coming from a J2EE dev background, I love the idea of Grails....all the benefits of using enterprise Java (Spring, Hibernate, etc. ), but with the simplicity of Rails (scaffolding, convention over configuration, Groovy scripting, etc.). My company is considering rolling out a new enterprise scale web application built on top of Grails. I heard there were some problems with stability and defects in earlier releases of the platform. However, I haven't seem many complaints since becoming part of Springsource.
I am curious if anyone has any opinions or experience in using Grails for large scale web apps. Is it stable? Reliable? Any gotchas I should look out for? Any other observations? Thanks!

Short Answer
Yes
Longer Answer
Stability and reliability of the core has improved considerably since 1.0. It seems that each new release introduces a few bugs, but workarounds/fixes tend to emerge pretty quickly
In terms of performance, Grails has always been adequate (for my purposes). Given that Grails is really a Groovy DSL for Spring MVC and Hibernate, and most of Grails is actually written in Java, it would be surprising if Grails' performance was substantially different from a Java web framework like Spring MVC.
The situation is very different when it comes to Grails plugins. Most of these are maintained by the community rather than the Grails team and are much more likely to be abandoned, or fail to keep up with newer versions of Grails. The officially supported plugins (e.g. Spring Security) are of course an exception to this statement.

For some previous discussion on this topic take a look at
Is Grails worth it? (from late 2008)
Is Grails (now) worth it? (from 2010)

I'd say yes, as long Grails 2.0+. With Grails 3.0 which is based on Spring Boot, you can improve developer productivity a lot more.
This Pivotal article justifies it.
http://blog.pivotal.io/pivotal/case-studies/netflix-amazon-asgard-groovy-grails-elastically-scaling-apps-for-40-million-internet-tv-viewers