Re: [Springframework-developer] Component Centric Web Framework for Spring

[Steven D. <ste...@gm...>]

Cristian,

I think the Java generation can be useful in the early stage of the
Wedge framework, but I'm with Seth in that the binding code should
eventually be autogenerated, preferably with a tool like ASM.

Wedge looks promising, and I would like to see a sample application
that demonstrates its features, like for example complex form
hanlding, validation, AJAX and so on.

Steven

On 3/25/06, Cristi PASCU <cri...@cr...> wrote:
>
> Thank you very much for your reply!
>
> When I started to work on 'wedge' my objective was to have a framework
> does not require either specific page class hierarchy or cluttered
> templates.
> The solution I came up with for achieving this was to have "the missing"
> code generated. First, I used javassist to create and load classes at
> runtime. I rapidly realized that this is not such a good idea because:
> 1. The generated code is invisible to the developer. Any exception that
> you get at runtime would be hard to debug if it occurred within this
> generated code. 2. There are a few limitations when using javassist. For
> instance, it does not supports inner or anonymous classes. 3. Class
> loading problems that may occur within a web container environment.
>
> Although I would like very much to come with a 'cutting [w]edge' web
> framework, I believe that it is more important to have that attitude
> that will keep us moving further. Wicket came up with a good change of
> view and perspective on java web development. As I studied some wicket
> example I was impressed by the simplicity of the xhtml templates and the
> elegancy of the java classes. Still, I try move things a little bit
> further. I want also to have clean java code, that is, to minimize the
> dependency of the application to the framework. In a wedge application,
> the framework takes your code and integrates it within its generated
> classes. In wicket applications you build your pages using specific
> framework components into the java code. That impose the creations of a
> lot of java objects, one for each component. For instance, take this
> example:
>
> In wicket:
>
> this.add(new Label("message", "Hello World!"));
>
> <span wicket:id=3D"message">Message goes here</span>
>
> This code must create a tree of components the must be synchronized with
> the template and rendered by passing through it recursivly.
> ---------------------
>
> In wedge:
>
> public String getMessage() {
>         return "Hello world!";
> }
>
> <span wid=3D"w:insert" value=3D"ognl:message">The message</span>
>
> This gets translated into:
>
> render("<span >");
> render(usrComp.getMessage());
> render("</span>");
> ---------------------
>
> As you can see, only a instance of the object behind the page is needed.
> The component (w:insert) is translated into code. All wedge components
> are translated into code and will not do any action at runtime.
>
> Embedding your code into the framework code frees your applications from
> compatibility breaking with future versions of wedge. If your
> application consists of pages that are simple POJO's than framework
> improvements for performance and clarity of the generated code will not
> affect the application code. And I think this is good thing. Backward
> compatibility is something that makes evolution of frameworks slower.
>
>
> Yes, you need to restart the application and regenerate the code as you
> make changes to the pages. But:
> 1. I use jetty as an embedded server which is very practical and rapid
> way to develop application. It does not take to much time to restart the
> server.
> 2. I've made the ANT task only for generating the pages. The generation
> requires all the application to be on the class path such that I start a
> new java process to do that. This takes time, about 5 sec for my 5 pages
> app. With the following code the time reduces to 1 sec.
>
>
> String base =3D "D:/PROJECTS/wedge/wedge.test.site/";
>
> WedgeCodeGenerator codeGenerator =3D WedgeCodeGenerator.getInstance();
> String webRoot =3D base + "web";
> codeGenerator.setWebRoot(webRoot);
>
> String sourceRoot =3D base + "wdg/src";
> codeGenerator.setSourceDir(sourceRoot);
>
> String wedgeApp =3D "wedge-application.xml";
> String appSpec =3D webRoot + "/WEB-INF/" + wedgeApp;
> codeGenerator.setWedgeApplication(appSpec);
>
> codeGenerator.generateFiles();
>
>
> 3. If I refresh the eclipse project folder, the eclipse compiler
> compiles the files much faster than ANT.
> 4. An eclipse plug-in would be quite helpful! (Note that the simplicity
> of the framework will make possible to have a powerful enough IDE
> plug-in)
>
>
> About defaults and conventions, I have been thinking to introduce them.
> But still, I need a way to figure out which spring beans are wedge
> components and which are not. More, you are not bound to spring, you can
> simply specify the class directly. For the file attribute, you can
> specify any kind of path that XmlWebApplicationContext can manage. The
> file name is not always suitable for being an id. Anyway, you are right,
> if is possible, that the developer should have the possibility to use
> this conventions.
> Unfortunately I do not see how I can make the wedge framework to work
> without any kind of information source. Anyway, the amount of xml will
> be kept to the minimum.
>
> Thank you very much for your interest and I will try that a.s.a.p to
> have a first release and more documentation.
>
> Help on developing is really needed! If there is anyone interested,
> please contact me!
>
> Cristian
>
>
>
> -----Original Message-----
> From: spr...@li...
> [mailto:spr...@li...] On Behalf
> Of Seth Ladd
>
> Sounds like an interesting project.  Can you compare to Wicket
> (http://wicket.sourceforge.net/) ?
>
> The one concern I have is, how quick is it to test changes to the code?
> From the impression I get, if I want to make a change to a page, I need
> to shut down the application and restart it so that the ANT task can
> recompile the page classes?
>
> IMHO, a really cutting edge Java web application framework would require
> absolutely zero shutdown and restarts of either the servlet container or
> the servlet context itself.  Do you happen to plan on attacking the
> "constant redeploy" problem?
>
> Also, not sure if it does already, but does the configuration support
> sensible defaults, or conventions?  A really cutting edge Java web app
> framework would be able to run with zero XML.
>
> For instance, given your example:
>
> <application>
>
>         <page id=3D"home" file=3D"home.html" springBean=3D"home" />
>
> </application>
>
> I would image the default rule would be: "convert home.html to 'home'
> and find the matching bean by name".  In other words, don't make me
> write the above XML when it's obvious what I am declaring. :)
>
> Seth
>
>
> On 3/24/06, Cristi PASCU <cri...@cr...> wrote:
> > Hi everybody!
> >
> > My name is Cristian Pascu and I am developing an opensource project
> > named "wedge" (http://wedge.sourceforge.net) which is a component
> > centric web framework built on Spring.
> > It is similar to Tapestry and aims to provide some features that are
> > not yet available in Tapestry. Namely:
> >         * better and easier integration with Spring.
> >         * simple POJO classes behind clean xhtml pages. There is no
> > need to extend specific framework classes. The classes are easy to
> > test as they are easy to instantiate and configure.
> >         * all page classes may be managed by Spring inversion of
> > control container. This way, services from the business tier will be
> > injected in the presentation tier.
> >         * a simpler component/pages specification. There should be a
> > single configuration file and not one for each component.
> >         * the configuration of the application should be very simple,
> > one xml file with few tags and attributes. Much of the configuration
> > will be on the spring side.
> >
> > Most of these features are already implemented. There is still some
> > work to do on the validation mechanism. The upload support is also
> > missing but is not to hard to implement.
> >
> > But what differentiate the wedge framework from tapestry framework is
> > that instead of using runtime reflection it follows the "code
> > generation" paradigm. Let me be more specific: The connection between
> > the template and the class behind it is made through an intermediary
> > generated class that binds the two. Thus, the generated code "knows"
> > exactly which is the structure of the template, that components to
> > render, what classes to use and what methods to call at the right
> time.
> > The generation is done with an ANT task each time before the
> > application is started.
> >
> > If you have any thoughts on what I am trying to do, critics or
> > advices, all are welcomed!
> >
> > Do you think that such a framework is needed?
> > Do you think that my approach is a good one?
> >
> > Thank you very much!
> >
> > Cristian
> >
> >
> > PS: There is not any release yet. You may check out the project from
> > sourceforge: https://svn.sourceforge.net/svnroot/wedge
> >
> >
> > -------------------------------------------------------
> > This SF.Net email is sponsored by xPML, a groundbreaking scripting
> > language that extends applications into web and mobile media. Attend
> > the live webcast and join the prime developer group breaking into this
> new coding territory!
> > http://sel.as-us.falkag.net/sel?cmdlnk&kid=110944&bid$1720&dat=121642
> > _______________________________________________
> > Springframework-developer mailing list
> > Spr...@li...
> > https://lists.sourceforge.net/lists/listinfo/springframework-developer
> >
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by xPML, a groundbreaking scripting
> language that extends applications into web and mobile media. Attend the
> live webcast and join the prime developer group breaking into this new
> coding territory!
> http://sel.as-us.falkag.net/sel?cmd=3Dk&kid=110944&bid$1720&dat=121642
> _______________________________________________
> Springframework-developer mailing list
> Spr...@li...
> https://lists.sourceforge.net/lists/listinfo/springframework-developer
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by xPML, a groundbreaking scripting langua=
ge
> that extends applications into web and mobile media. Attend the live webc=
ast
> and join the prime developer group breaking into this new coding territor=
y!
> http://sel.as-us.falkag.net/sel?cmdlnk&kid=110944&bid$1720&dat=121642
> _______________________________________________
> Springframework-developer mailing list
> Spr...@li...
> https://lists.sourceforge.net/lists/listinfo/springframework-developer
>
>