I have got my hands dirty in Spring Roo, I have managed to do the persistence set up for Oracle Database. Although I have gone through the Roo documentation
but it not enough for me to understand the working of this framework.
I'm not aware of Spring-MVC ,Spring-Web Flow and AspectJ rather I have knowledge of frameworks such Struts, Tapestry and GWT. Could this be a hindrances in learning Spring Roo.
Please advice as I'm running out of patience of hacking the generated code by Spring Roo. Is it good idea for me to jump in a Spring Roo project?
The Spring Roo is more than a code generator. If you make changes in your Entities, it will change the codes automatically. More like a stateful code generator. First you need to find out the commands available in Spring Roo to configure a project.
You can configure which database you want to use, which orm you want to use, whether you want to use unit-tests or which view technology you want to use. Give the commands and Roo takes care of it. You don't need to worry about which libraries to download, how to make pom file for Maven.
The configurations you got using Spring Roo is quite a standard one. I used it in the previous projects. The code generated is not optimal, for example , you can use generic daos instead of Roo generated DAO if you consider the generated codes are verbose.
Another example says, in Flex, you can use Roo to generate configurations for the communication between Flex client and J2EE server. For my case, I use parsley and observer patterns for flex client instead of the codes generated by Spring Roo. But by looking at the codes what the Roo-Flex team wrote taught me quite a lot of things too. The same things goes for GWT and Spring MVC.
So what am I saying is that as the programmers, we need not only to code we also need to read what the other wrote too . I strongly believe that it is a good way to learn and share best practices. You might get some boilerpate codes, but if you know how to apply design patterns or can enhance on them, those codes will become the gems.
Learning curve ? Not at all. The commands are as simple as DOS commands.
Spring ROO is just a command line interface for generating a lot of the boiler plate code you usually need to write to get things done in java projects.
If you want to add GWT into your project, just type "gwt setup" after you've created your models and you have full-on GWT support in your project.
The default view being generated in Roo uses Apache tiles, similar to Struts, Tapestry and JSF.
If you want to use web flow, just type in "web flow" and it'll be integrated for you, for struts support, you'll need a plugin, I've seen a couple of them floating around the web, haven't tested any of them; same goes for tapestry, you can either use a tapestry plugin or do the integration by hand.
AspectJ in the context of Roo generated scaffold is used to hide all the boilerplate code that makes code look like spaghetti, you can move code out of the aspectJ generated files into your own java files and modify them if you need to.
As for the hacking, it takes a lot of patience and time to fully understand where you should change stuff and where you shouldn't, all I can say is, happy hacking :-)
springroo is based on many technologies, but all what springroo does is to help you to generate scaffolding code using lots of addons you can install
at least it's my view of this framework after three days of learning it
Related
I am going to develop small ERP System. The product has only one major requirement to support multiple databases.
I have planned to use
Front side: JSP+JSTL+JQuery [I have good command on that] + I have create my own custom component for re-useability and full control of my component.
Back-end: Using hibernate[ORM] framework [due to Cross database] and I have also good knowledge of it.
I am happy with above and feeling confident to build product, soon.
BUT
My friend made me unhappy ;)
My friend suggest me that you should use spring MVC with hibernate, because when your product become large,it create a problem in future due to many developers involvement.So, you should use framework , so that every developer follow the pattern and then your product remain stable.
But I am already follow hibernate pattern :)
I have no any knowledge of Spring MVC. When I search about spring MVC , I found it is a different thing than my requirement and also find Cross database context issue, if I use Spring MVC with hibernate.
Either should I used spring MVC or not. If yes, I have to learn then Spring MVC and
I do not want to waste my time, if spring is not suitable for me.
That's why I am consulting that forum , to go for right direction.
Actually, Spring MVC does not interfere with any of the components you have chosen already. It just introduces a way to structure your frontend - backend interface in a standard way (MVC pattern).
I actually run the very same setup and is very happy with it. (Jquery, JSP, Spring MVC, Hibernate).
Other than structure, you will get easy return of invested time in Spring MVC when you want to do Ajax heavy things (as I assume you want in an ERP app). It's trivial to build AJAX/JSON stuff with jQuery and SpringMVC.
Using Spring MVC will make your application structured. This will help in future maintenance.
Your code will look clean and you can separate your code into components and Spring will help you wire them together and you might decide to combine them in different ways, or wish to make it easy to swap out one component for another depending on different settings or environments. This would be great for multiple databases when used with hibernate.
Learning spring will not take much time if you have a good knowledge of Java EE.
I just got a book from Apress (Spring Recipes), and I started to learn the Spring framework and now I have a problem. I followed the steps in the book to add the Spring .jar-s to the build path. Everything seems to work fine, but the code I have just doesn't work.
My question is that: do you know about a good tutorial for integrating Spring with Eclipse? I googled for a while and I only found old tutorials (from 2007-2009) which just doesn't work with Eclipse 3.6.
use STS - Spring Tool Suite. It is built-upon Eclipse IDE. http://www.springsource.com/developer/sts
STS helps hugely (you can get it from the Eclipse Marketplace for free). The features of it I particularly like? Here's a few:
As a user of XML-based configuration I love having the bean graph visualized – it's great for reports – but the content assist when filling out bean properties is invaluable, both for names and for values (i.e., it makes writing that much XML not suck too much). I imagine that if you're heavily using autowiring or Java configuration it's less useful.
Refactoring support is also good (these days; it used to have bad bugs so you want to be patched up to date) and I've found that the pointcut search to be the only way to write a complex aspect without lots of round-trip testing.
But not all of STS is perfect. In particular, some of the content assist modes don't (seem to) help (me) that much, and it really only works well if you have your properties defined using public setters. Still, it helps a lot…
Apart from using STS as suggested, you can use spring and eclipse without any integration. There isn't that much an IDE integration can offer anyway (apart from autocompletion in the xml, and linking the xml to classes (inclding aop))
After many problems, I have finally set up Roo with a simple example.
Now I just want to know how far should one go with Roo.
I mean likes fields, classes, everything can be done through the Roo shell.
Now, for example, I have 15 classes in my project. I'm confused - to what point should I make classes with the Roo shell and when should I leave Roo and start working like normal?
Also, Roo has its own GUI and layout design. Can we change that also?
You can also try Telosys Tools, it has been design to generate a full web application from an existing database.
It’s more flexible than Spring Roo, you can customize the model and the generator templates
The tutorial for code generation with Spring MVC and Spring Data is here : https://sites.google.com/site/telosystutorial/
I personally use ROO to generate the models, controllers and the scaffold, then I strip out a large chunk of the Scaffold so that I can put my own components in there and some of the controllers I still need to write by hand.
When you reach the point where you have an interface generated by ROO, start hacking away and only use ROO to add new domain models.
I agree with Jan. The Roo Scaffold application is just that: a structure that you can use to build your own application around, but that isn't something you'd like to expose as a client in its own right. So far I have used Roo when creating or changing the data model, and have allowed it to manage and modify a lot of the code that is necessary for interacting with the data model. But any work I'm doing outside of the model I do myself.
My advice: use Roo where its strength is, and then do the rest by hand.
I'm currently using Java & Spring (MVC) to create a webapp, and I'm considering moving to Grails. I'd appreciate feedback/insight on the following:
I have multiple application contexts in the current Java/Spring webapp that I load through the web.xml ContextLoaderListener; is it possible to have multiple application contexts in Grails? If, yes, how?
This webapp extensively uses a CXF restful web service and the current Java/Spring webapp uses the bundled CXF HTTP client. Can I continue to use the (Java) CXF HTTP Client in Grails?
I implemented Spring Security using a custom implementation of UserDetails and UserDetailsService, can I re-use these implementations in Grails "as is" or must I re-implement them?
There is an instance where I've relied on Spring's jdbc template (rather than the available ORM) and an additional data source I defined in app context, can I re-use this in Grails?
I plan on using Maven as the project management tool; are there any issues of using Maven with Grails where there is a combination of groovy and java?
Edit:
I'm considering moving to Grails to make the development of the web component of the webapp "faster," a la Ruby-on-Rails. Also, I'm considering Grails rather than say Ruby-on-Rails, because I want to continue to use the JVM and I've dabbled with Grails in the past and it was fairly easy to pick-up and use.
Probably. Grails uses a sub-class of Spring's ContextLoaderListener class which it configures in the web.xml file. I can answer more precisely if you let me know how you do it with Spring MVC.
Yes. You might even be interested in the CXF plugin, although I can't vouch for it:
http://grails.org/plugin/cxf
You should be able to use them as-is. However, you might want to check whether this is easily done with the Spring Security plugin. I believe it is, but you'll be able to get a definitive answer from Burt Beckwith, the author of the plugin.
Yes. You can also get hold of the Hibernate session factory to do raw Hibernate stuff. GORM can also work with multiple data sources:
http://grails.org/plugin/datasources
Another Burt Beckwith one :)
It depends on what you mean by "a combination of Groovy and Java". You can build Grails projects with Maven, but the integration isn't entirely smooth. If you have Java and Groovy in your Grails project, then that's taken care of automatically.
In response to Bozho, I use standard Grails services + GORM and wouldn't do it any other way. Note that if you use Java for services and the domain model, you won't have automatic reloading of services. You also lose the benefits of expressiveness and conciseness that Groovy bring.
If you want, you can use static types in Grails services to make it easier for your IDE to provide code completion. It can also give you hints on properties and methods it doesn't recognise (which would corresponding to Java compilation errors). That said, even if you use static types, Groovy can't do type checks at compilation time. You'll only find out about them at runtime.
You can do all these things in grails. It supports all existing Java classes and spring configurations (grails is built ontop of spring mvc)
However, I really wouldn't recommend moving the whole application to grails. You can perhaps move only the web layer, if you have web developers that are not java experts.
The service layer, the data access, etc, better remain pure Java. That is, only your web controllers - the components that gather the user input, handle http requests and sessions, should use grails. The rest - the stateless service classes and your domain model would better be Java. That's my opinion, but I have already some experience with grails, and static typing in the service layer will save you much trouble.
2) Yes you can use CXF as is. There is a nice layer on top of CXF called GroovyWS. I have only used it for consuming SOAP services, but maybe it has something for REST as well. It's really easy to use.
For consuming REST services I have used HTTP Builder
4) Yes. You can continue to use e.g. spring config for configuring the datasource, or any other way you do it today. Multiple datasources is no problem.
5) I have recently tried using Grails (1.2.1) with Maven. It works, but there has been some issues with both Maven and Grails trying to do dependency management. The documentation is maybe the worst part. I haven't tried upgrading to 1.3 yet because of some major Maven-related JIRAs, but 1.3.2 is right around the corner, and those issues have now been resolved :) There will also be a 1.3.2 maven archetype. Looking forward to that. "Deployment and resolution of plugins from Maven repositories" is one of the new features of Grails 1.3, so things are probably better. Roadmap for 1.3.2 says release today, but there are 8 issues left at time speaking, so my guess would be tomorrow, the Grails releases are usually on time. If you can wait for that, you will probably save yourself some trouble.
If you are looking for rapid application development but aren't otherwise particularly enthused about groovy, you should look into spring-roo. It offers the same kind of RAD functionality, but builds a completely standard java + ORM + spring-mvc app (which has no actual dependencies (runtime or compile) on roo). It's definitely not as mature as grails, but you may find that it better suits your existing experience with statically typed java code and existing ORM, etc. I've only done a couple of small pet projects in roo, but I've been very impressed so far, particularly with how easy it is to customize the generated code and move back and forth between written and generated code. The initial tutorial is very rapid and quite revealing.
Appfuse vs. Roo, what would you use and why?
What are the sweet spots of each.
As per the answer I gave to TheServerSide thread on this issue...:
AppFuse aims to provide a single initial scaffold of your new project. This is similar to Maven archetypes or Eclipse's "new project" features in that you run them once at the start of a new project and then you maintain the scaffolded code going forward. The scaffold system has no further involvement in your project once you've run it once.
Roo, on the other hand, provides a round-trip aware active code generator for your long-term usage on a given project. As such Roo offers value both at initial creation time as well as whenever you are modifying the project going forward.
In practical terms this means as you evolve your project, Roo will automatically maintain certain files. To take a simple example, when you add (or remove) a field, Roo will update the toString, getters/setters, JSP pages etc for you automatically. It also offers commands so you can add new capabilities later. So if you need to add security six months after you created the project, you just "security setup". Or if you need to send emails, you just "http://static.springsource.org/spring-roo/reference/html/command-index.html#command-index-email-sender-setup". There are similar commands for many other capability areas well, such as Spring Web Flow, JUnit, Selenium, common JPA providers etc. You just defer the decision as long as you like, and Roo will only add those capabilities at the time you ask for them (and it will also automatically use those new capabilities in your project).
There are many other differences as well. Roo allows extension via user-developed add-ons, it offers a highly usable shell, it allows you to incrementally build a new project and add features only when required, it extensively supports the latest versions of the major Spring technologies, it comes with a SpringSource-developed (and therefore endorsed) application architecture and so on.
A read of the Roo Reference Guide's Introduction Chapter or simply completing the ten minute test project will illustrate they are very different in approach.
My notes on AppFuse and Roo:
AppFuse
Is a fully working template application/project.
Traditional DAO <-> Service <-> Controller architecture
Easy to get started with maven archetypes
Great documentation and tutorials
Not really up to date. Spring 3 final is soon to be released and AppFuse is based on 2.5 (?)
Spring Roo
Spring Roo on the other hand is a tool that speeds up development by using code generation.
Getting started with a new, fully configured project takes 1 minute
Creates rich domain objects where CRUD are weaved into the domain objects using AOP instead of traditional DAOs/services
Hard to grasp if you are new to Spring
Documentation is not that good yet
Really cool! I.e. add Spring Security to your project with just one line of code!
Telosys (a lightweight code generator) is also a good alternative. See
http://www.telosys.org/
It produces very clean code (without adherence like ApectJ)
and the templates are customizable
There's a stack of templates designed to generate Spring MVC web apps (and many others to generate code for other kinds of frameworks).
for me Appfuse but it is not up to date, but the spring roo using aspectj and there are parts of code you should not touch and I do not like that.
Spring Roo
Pros.
1.
Customizable : You add and remove diffrent framework and addon as per your requirement.
Database Reverse Enginnering : Create CRUD applications if you have database schema ready.
Strong Spring community support.
NOSQL MongoDB support
Can create required add-ons.
Cons :
Require in depth knowledge of Aspect oriented and Spring stack.
Require little more time to learn spring roo as compared appfuse.
Appfuse:
Pros:
Good one to start small and mid-size enterprise application with
struts, JSF and Spring
Complete open source code.
Enough documentation.
Twitter-bootstrap ready.
Cons:
Customize application but not like Spring roo.
For starters, roo looks more over engineered, with code generation, use of aspect oriented programming and more.
Appfuse doesn't seem to be maintained anymore seeming that the last version was released on May, 2008.
Roo it's right now a little bleeding edge, because of the use of the still-unreleased version 3 of Spring Framework, but that will change, and that version brings a lot of interesting changes to the table.
It also upsells you on more of the Spring technologies portfolio, such as STS and tcServer, and makes it dead easy to use Spring Security and Spring WebFlow.
I am going with ROO.
I am already using
Spring
Spring ORM/JDBC
Spring MVC
Spring Remoting
STS (Tool-suite)
So my preference is SpringSource products, AS I am already familiar with API style, documentation, conventions of SpringSource and even their coding practices once I extended/implemented security-framework's code.. ;-)
so, my advice is go with the tool/framework which is more natural to you..
Cheers,
AppFuse -> change to SpringFuse
I more prefer use SpringFuse
The question is some years old and in the meanwhile there are new productivity tools I want to point out:
Generjee. Generjee is a full-online tool. You define online your requirements and the (optional) data model. Then you get the generated code as a download.
The generated code is independent from generjee. The tool is useful for starting development projects from an integrated full-stack code base.
generjee will generate for you:
JPA code according to your data model
JSF code to create, read, edit, filter, sort and export data
user management, registration and login
your defined user roles and specific access permissions for this roles
I18N support
file upload support
Forge. To describe it in some words, it is "like Roo". But Forge is not as strong based on Spring and AspectJ as Roo.
AppFuse has integration with Tapestry, Wicket and other Web frameworks which Roo does not - yet
The AppFuse project was shut down in April 2016. Its founder, Matt Raible, recommends using JHipster as an alternative.