Angel “Java” Lopez on Blog

Weeks ago, I was working on my open source project AjTalk, an Smalltalk-like interpreter, written in .NET (using C#), and now, I want to present the status of that work. For years, I was interested in Smalltalk development, altought only in my free time, never as a professional developer. It’s not the first time I wrote such kind of interpreter (my first one was written in the eighties, and it was very simple: no garbage collector, only text commands), but this time I feel it could become a very complete implementation.

Current version is minimal, but it is taken form. The idea is to have dynamic objects, like in Smalltalk, and, at some point, add prototypes a la Self. The objects and the interpreter would access full .NET framework and other libraries, as I did in my AjBasic interpreter (included as part of AjGenesis code generation project).

The very initial version is now published at Google Code:

http://code.google.com/p/ajtalk/

The solution

I’m using Professional VS 2005. There are four projects in the solution.

AjTalk is the main project, a class library containing the core of the system.

AjTalk.Test01 and AjTalk.Test02 are console applications, only for manual tests.

AjTalk.Tests contains the unit tests, written using NUnit framework 2.2.8.

Most of the core system consists of interfaces, defining the base behaviors, and classes, implementing such interfaces.

The object

The base is to have an interface to represent any object:

using System;

namespace AjTalk
{
    public interface IObject
    {
        IClass Class { get; }
        object this[int n] { get; set;}
        object SendMessage(string msgname, object [] args);
    }
}

I could make a message an object (of type Message), but for now, the message is only a name, and an array of arguments.

The index this[int n] access the intance variables. Each value in AjTalk can point to any .NET object, not only the ones that implements IObject. In this way, I can manage int, long, String, DataSet, from other IObject objects.

The class

I’m implementing a simple IClass interface, without distinguing between class, behaviours, and other classes, as in the classic Smalltalk. I’ll separate these classes in a future version. For now, I’m managing only a interface:

using System;
using System.Collections;
using System.Collections.Generic;

namespace AjTalk
{
    public interface IClass : IObject
    {
        IClass SuperClass { get; }
        string Name { get; }

        void DefineClassMethod(IMethod method);
        void DefineInstanceMethod(IMethod method);

        void DefineClassVariable(string varname);
        void DefineInstanceVariable(string varname);

        IObject NewObject();

        IMethod GetClassMethod(string mthname);
        IMethod GetInstanceMethod(string mthname);
        int GetClassVariableOffset(string varname);
        int GetInstanceVariableOffset(string varname);
    }
}

There is a dictionary of class and instance methods, and lists of class and instance variable names. It doesn’t have support for indexed variables, yet. IClass interface is now implemented in a BaseClass class.

The method

There is an interface implemented in Method class:

using System;

namespace AjTalk
{
    public interface IMethod
    {
        string Name { get; }
        IClass Class { get; }
        object Execute(IObject receiver, object [] args);
    }
}

The concrete Method class, has an Execute method:

        public object Execute(IObject receiver, object[] args)
        {
            return (new ExecutionBlock(receiver,receiver,this,args)).Execute();
        }

Execution blocks have local variables and arguments. The Execute of an execution block takes the instructions (bytecodes) from “compiled” methods, and then it executes them. Here, I took a departure from Smalltalk ideas: the execution block is not an AjTalk object. In this way, I could run this interpreter without implementing a lot of base classes. I have to research the advantages and problems that this decision could have in the overall design and implementation.

The bytecodes

I must think about using a tree of objects (as in Interpreter pattern) instead of bytecodes. But in this version, bytecodes are used. These are the basic instruction that my “virtual machine” understands and executes step by step.

There is a bytecode list (an enumeration)

namespace AjTalk
{
    public enum ByteCode : byte
    {
        Nop = 0,
        GetVariable = 1,
        SetVariable = 2,
        GetArgument = 3,
        SetArgument = 4,
        GetConstant = 5,
        GetLocal = 6,
        SetLocal = 7,
        GetClassVariable = 8,
        SetClassVariable = 9,

        GetSelf = 20,
        GetClass = 21,
        GetSuperClass = 22,
        NewObject = 23,
        Pop = 24,
        ReturnSub = 25,
        ReturnPop = 26,

        Add = 40,
        Substract = 41,
        Multiply = 42,
        Divide = 43,

        Send = 50
    }
}

The bytecodes contained in a method, are interpreted and executed by the execution block. An excerpt of that code:

    while (ip < method.ByteCodes.Length)
    {
        ByteCode bc = (ByteCode) method.ByteCodes[ip];
        Byte arg;

        switch (bc)
        {
            case ByteCode.ReturnSub:
                return null;
            case ByteCode.ReturnPop:
                return Top;
            case ByteCode.GetConstant:
                ip++;
                arg = method.ByteCodes[ip];
                Push(method.GetConstant(arg));
                break;
            case ByteCode.GetArgument:
                ip++;
                arg = method.ByteCodes[ip];
                Push(arguments[arg]);
                break;
....

Test everywhere

The initial code was written in VB.NET. Last year I began to rewrote the original source code to C#. Then, this year I switched to “TDD mind”, so, late but sure, I added NUnit tests:

Bootstraping

I plan to use a text file, with an ad-hoc format, to inject the definitions of the initial classes and objects. Now, in the AjTest.Test02, there is an example of such format in Definitions.txt file:

class Point
variables x y

method
x ^x.

method
y ^y.

class Rectangle
variables point1 point2
class Square Rectangle

Next steps

There is a lot of work to do:

- Complete the hierarchy of base classes (Behaviour, Class, ….)

- More byte codes

- Support of local variables in methods

- Standard file text format

- Access to native .NET objects

- Use from .NET applications

- Define the classes and methods for a minimal implementation

- Serialization/deserialization of memory image

- Support for adding variables to a class with created instances (this is a tough problem)

- Support for become:

- And much more….

But I’m confident on the shape the project is taken. I’m applying “baby steps”, to improve the base code and functionality.

Angel “Java” Lopez
http://www.ajlopez.com/en

I’m back! I want to write in this post about the new features I’ve added to my code generation project AjGenesis. The main points are:

- A new interactive way to use it, a web based application

- Examples with that new interface, to generate C#, VB.Net, Java and PHP applications, all from the same abstract model

These features were published last January at release 0.5 in CodePlex

Using AjGenesis

Years ago, I began to work in this project. Those days, the only way to execute and use it were the console, with commands. Later, I adopted NAnt, and NAntGui, as interfaces to launch the system and execute tasks and transformations. The project can be used as a set of .DLLs, referenced from custom application. Carlos Marcelo Santos has published posts (in Spanish) about how to use AjGenesis from NAnt

Cómo generar código con AjGenesis sirviéndonos de NAnt
Cómo generar código con AjGenesis sirviéndonos de NAnt - Parte II

Last year, 2007, Jonathan Cisneros wrote a great Windows interface, named AjGenesis Studio:

AjGenesis Studio: una IDE para AjGenesis
AjGenesis Studio at CodePlex

As an exercise at the end of year 2007, I wrote a new project in the original solution, and I published last January inside the last release 0.5 at CodePlex.

It’s a new web project, that I called, inspired by the work of Jonathan, AjGenesis Web Studio

The main point is, with this project, to show that the functionality from the AjGenesis core can be used and exposed in our applications, this time from a web interface, using ASP.NET. With this approach, we can put a server in our internal network or at Internet, and create, edit, upload and download user defined models, generate source code, and download the results. We could call this, Code Generation as a Service.

Let’s look the project, its functions and some inners.

The project

It’s a web project, named AjGenesis.WebStudio.

To launch it, load the AjGenesis.sln solution (located at src directory inside the version 0.5 from CodePlex) with Visual Studio 2005, and build the complete solution.

(If you have errors in the AjGenesis.WebStudio project, probably you must add a reference to a .DLL that manages the file compressions. Add as reference to the project the file src\Libraries\Ionic.Utils.Zip.dll. This library generates .zip files, and expand them).

In the web solution, I used Master Pages, and a default Theme. It doesn’t manage a database, only files and directories. We’ll see below that the project has a mini wiki, to write help pages that can be defined by the user.

The look and feel is simple, but you can change the theme and master page to obtain a better result.

The system can generate code in the fly, and download it as a .zip file. You can use the classic AjGenesis system, as I explained in past post, but now, we can use it from any browser. You don’t need .NET in your client machine, and you can work from other operating systems.

After the build of the solution, click right mouse button on the file \Default.aspx in project AjGenesis.WebStudio, and choose View in browser…:

This is preferred to launch the project in debug mode: if you debug the project, it will execute the code generation task in very slow way.

Working Directory

The system uses a working directory, with a structure that resembles the examples at AjGenesisExamples3.zip (to download from examples at CodePlex). The default working directory is AjGenesis.WebStudioWorkingDirectory:

Some directories:

Projects: There is a directory here for each project model.
Templates: Containing the template files to process during the code generation phase
Build: The code generated code resides in this directory.

The working directory can be changed using the left menu option Working Directory and then, choosing the option Change Working Directory:

Generating Hello World

As I used to comment in my speeches, the acid test of all code generation system is to help us to model and produce the most simple and universal software system, the reknowed Hello World example. There is an explanation on how to do this with classic AjGenesis in my post

Code Generation with AjGenesis- A Hello World application

In AjGenesis Web Studio, select the left menu option Generate. This is the result page

There are two dropdown list. We’ll see how they are filled. At the first list, there are the defined projects:

Choose HelloWorld. Then, go to technology list, showing:

In this project, only Java and Net20 are the defined technologies. Other projects have more technologies. Choose Java and the press the Generate button. After a while, the page shows a result.

How this works?

First point, the system manage a working directory, the default is AjGenesis.WebStudioWorkDirectory, a syster directory of the web project. There is a Projects project, cointaining more directory, each one is a project for the system.

The directory AjGenesis.WebStudioWorkDirectory\Projects\HelloWorld has a Project.xml file that describes the free model we use in this case:

<Project>
  <Name>HelloWorld</Name>
  <Description>HelloWorld Example</Description>
  <GenerateTask>GenerateHelloWorld</GenerateTask>
  <Message>Hello World</Message>
</Project>

Inside the directory project, there is a Technologies directory. Each XML file that is found there it’s managed as a technology model in the system. Again, as I explained in my posts on AjGenesis , this file has a free model, you can put here anything to process from code generation process. The concepts of project and technology is not mandatory, but AjGenesis Web Studio takes them as special directories, and read them to obtain the list of project and associated technologies.

The technology Java.xml file contains:

<Technology>
  <Name>Java</Name>
</Technology>

and Net20.xml contains:

<Technology>
  <Name>Net20</Name>
</Technology>

Nothing special. The value Name inside Technology drives what steps to execute during code generation phase.

After press the generate button, the system loads the project and technology model in memory. Then, it executes the file GenerateCode.ajg located in directory Tasks. But if the project file contains a  GenerateTask tag, as in this project, that file is executed instead the default one. This is the content of Tasks\GenerateHelloWorld.ajg:

PrintLine "Generating HelloWorld..."

if not Project.BuildDir then
    Project.BuildDir = "${WorkingDir}Build/${Project.Name}"
end if

if not Project.Title then
    Project.Title = Project.Name
end if

if not Project.Version then
    Project.Version = "1.0.*"
end if

if not Project.Language then
    Project.Language = "en"
end if

if not Project.SystemName then
    Project.SystemName = Project.Name
end if

PrintLine "Creating Directory ${Project.BuildDir}"

FileManager.CreateDirectory("${Project.BuildDir}")

Include "Tasks\BuildHelloWorld${Technology.Name}.ajg"

This file is written in an scripting language, a pillar of AjGenesis, affectously named AjBasic. The Project variable is the representation in memory of the tag <Project> loaded from the Project.xml file, described above. In memory, it is as an object, with properties. It’s not an static object: you can add properties in runtime, as Project.Version in the above text. If in an if sentence we asks the value of  Project.Version and this doesn’t exist, it’s not an error, the result is simply Nothing and that value is treated as false for boolean evaluations (like in other languages, notably PHP).

Finally, the last line includes an additional file to process dinamically. The string contants in AjBasic have expression expansion: all between ${ and } is evaluated. Using the value of Technology.Name (obtained from loading in memory the XML technology file), it executes BuildHelloWorldJava.ajg or BuildHelloWorldNet20.ajg. The first one contains the steps to generate a .java file:

<#

PrintLine "Generating Solution ${Project.Name}"

if not Project.Title then
    Project.Title = Project.Name
end if

if not Project.Version then
    Project.Version = "1.0.*"
end if

PrintLine "Creating Directories"

FileManager.CreateDirectory("${Project.BuildDir}")
FileManager.CreateDirectory("${Project.BuildDir}\${Technology.Name}")

TransformerManager.Transform("Templates\HelloWorld\ClassJava.tpl", "${Project.BuildDir}\${Technology.Name}\HelloWorld.java", Environment)

#>

This code uses inner AjGenesis utilities that create directories and expand a template file. ClassJava.tpl contains:

//
// Automatically generated by AjGenesis
// http://www.ajlopez.com/ajgenesis
//

public class HelloWorld {

    public static void main(String[] args) {
        System.out.println("${Project.Message}");
    }

}

All this produces the directory Build\HelloWorld\Java:

If we choose the Net20 technology, more files would be generated:

This is a more complete result. It has a VS 2005 solution file, two project directories, one in VB.NET and the other in C#, with source code and project files:

To see the generated code, its files and directories, go to left menu option Builds:

There is a directory for each project. You can download a complete directory, as a .zip file. We can create a file, download a directory, upload and expand a .zip file, and consult a help page. We can read, edit, and delete each file, in this case readme.txt.

Go to the View at HelloWorld directory:

You can browse directories, upload and download files. You can move a directory, or make a copy, to experiment without loosing the original content.

Generating C# and VB.NET solutions

After this simple example, let’s go for more. For more information about the next examples and their models, read:

Application Generation using AjGenesis

Go again to left menu Generate option. Now, we choose a project, AjSecondExample. Its technology list is more complete:

Choose CSharp2 and generate. This process creates a complete solution inside Builds\AjSecondExample\CSharp2. There is an Sql directory with DDL script to create a MS SQL Server database. Inside the Src directory, you’ll find the solution, composed of a set of projects:

This projects follows a layered archictecture, that you can modify from the templates. They use a data access layer, based on my AjFramework project, but you can change to access Enterprise Library or your own data services.

You must create the database using the generated scripts, and modify the connection string in the web.config to match your environment. The application can be compiled and executed:

The administration page shows:

Back to generate page, execute the VbNet2 technology, obtaining a directory with the same structure, now with Visual Basic.NET code:

All these depend of the specified technology inside the files at Projects\AjSecondExample\Technologies. The content of VbNet2.xml is:

<Technology>
    <Programming>
        <Dialect>VbNet2</Dialect>
    </Programming>
    <Database>
        <Dialect>MsSql</Dialect>
        <Name>AjSecondExample</Name>
        <Username>sa</Username>
        <Prefix>ajse_</Prefix>
        <Host>(local)</Host>
    </Database>
</Technology>

and CSharp2.xml:

<Technology>
    <Programming>
        <Dialect>CSharp2</Dialect>
    </Programming>
    <Database>
        <Dialect>MsSql</Dialect>
        <Name>AjSecondExample</Name>
        <Username>sa</Username>
        <Prefix>ajse_</Prefix>
        <Host>(local)</Host>
    </Database>
</Technology>

These data are used during the code generation process.

Note: inside Tasks\BuildTechnology.ajg file, you’ll find a new variable, injected from the web project:

if not Project.BuildDir then
    Project.BuildDir = "${WorkingDir}Build/${Project.Name}/${Technology.Programming.Dialect}"
end if

This is the WorkingDir variable that points to the working directory we selected in AjGenesis Web Studio.

Genarating Java and JSP applications

Many projects (AjFirstExample, AjSecondExample, AjTest….) can be generated in Java with JavaServer Pages. A build.xml file is generated to be modified and used to compile the generated code using the ant utility (you can use this file from an IDE, like Eclipse or NetBeans).

The target build  makes a .war deploy file. You can browse and modify the build.xml file to install the .war in a Tomcat web container.

The generated code is composed of .jsp files (web pages in Java) and Java code to compile that implements a service layer and a data access layer, exchanging simple JavaBeans between layers. The DDL script, in these examples, is a MySql one.

If you build and deploy the application, and install the database in a MySql server, you can see it working fromn a Tomcat server:

Generating application using Domain-Driven Design ideas and Hibernate, NHibernate

Many example projects (AjFirstExample, AjSecondExample, AjTest…) have defined technologies CSharp2DDDNh, JavaDDDHb, VbNet2DDDNh, and even VbNet2Nh. The DDD ones implement some concepts from Domain-Driven Desigg, as Application, Domain and Infrastructure layer, but the templates could be improved. The *Nh projects generate code for entities using NHibernate mapping. The *Hb projects use Hibernate.

An example of NHibernate generated mapping:

<?xml version="1.0" encoding="utf-8" ?>
<hibernate-mapping xmlns="urn:nhibernate-mapping-2.0">
    <class name="AjFirstExample.Domain.Customer, AjFirstExample.Domain" table="ajfe_customers">
        <id name="Id" column="Id" type="Int32" unsaved-value="0">
            <generator class="native"/>
        </id>
        <property column="Name" type="String" name="Name" length="255"/>
        <property column="Address" type="String" name="Address" length="255"/>
        <property column="Notes" type="String" name="Notes" length="255"/>
    </class>
</hibernate-mapping>

These solution can be built as the previous examples. You’ll see better generator templates for NHibernate in the AjOmar example at CodePlex. Post (in Spanish) explaining the process at

Generando código para NHibernate (Parte 1)
Generando código para NHibernate (Parte 2)
Generando código para NHibernate (Parte 3)

Generating PHP applications

If you choose PHP technology, web page with PHP code are generated, using MySql as database. The generated code can be copied to a directory in a web site that supports PHP, Apache, or IIS. AjTest example running in a IIS with PHP4:

Other directories

In the working directory, there are other directories:

SourceCode: Containing some source code that is copied into the solution during code generation.

Libraries: Additional .dlls, to copy and use in the generated applications.

The larger subdirectory  is Templates:

In previous posts I described in detail the creation and use of a template file. This is the content of a template that generate a C# entity:

<#

rem        Entity Generator
rem        for C Sharp


include    "Templates/CSharp2/CSharpFunctions.tpl"
message    "Processing Entity ${Entity.Name}"

include "Templates/CSharp2/Prologue.tpl"    

#>

/*
 *    Project ${Project.Name}
 *        ${Project.Description}
 *    Entity    ${Entity.Name}
 *        ${Entity.Description}
 *
 */

using System;

namespace ${Project.SystemName}.Entities {

    public class ${Entity.Name} {

//    Private Fields

<#
for each Property in Entity.Properties
    message    "Processing Field ${Property.Name}" #>
        private ${CSharpType(Property)} ${CSharpFieldName(Property)};
<# end for #>

//    Default Constructor

        public ${Entity.Name}() {
        }

//    Public Properties

<#
for each Property in Entity.Properties
    message    "Processing Property ${Property.Name}"
#>
        public ${CSharpType(Property)} ${Property.Name}
        {
            get {
                return ${CSharpFieldName(Property)};
            }
            set {
                ${CSharpFieldName(Property)} = value;
            }
        }

<#
end for
#>

    }

}

Help pages

I’m testing some ideas in this web project. There are code to create “wiki” pages, that can be used as help pages in the system.

You can edit this kind of pages:

There is a way to insert links to other wiki pages and to external ones. The page data is represented as a .NET object in memory, and that object is serialized in an XML file in Pages application directory. There is no database to store page: each one is an object, serialized in a .xml file.

Conclusions

I hope this system helps to a better understanding of AjGenesis project features and potential. With a more user friendly interface, the project can be used by more people. We can generate code from command line, or using NAnt utility, or AjGenesis Studio.

I ever remark a point: all these are examples, AjGenesis is not limited to generate only this kind application. You are in charge: you can define the model that matches your project requirements, using any technology you choose. You can codify DDD ideas in a different and better way. You can generate Hibernate/NHibernate mappins using other patterns and mapping idioms. You can use other ORM utilities, and generate code for them. You can generate PHP5 pages that use Prado, instead of plain PHP4. You can generate stored procedures for Oracle, instead of my examples that use MS SQL Server. Or write better ones. You can generate WinForms code, instead of ASP.NET page. You can generate JavaServer Faces pages, views, beans, and abandon JavaServer Pages. You can generate the code or text you want: e.g., unit tests, or documentation pages. I recommend that the generated code resembles the code you could generate manually.

I invite anyone that uses this system, to post his/her work, or leave comments here. You can participate in the Spanish list:

http://groups.google.com/group/codegeneration?hl=es

Next steps

Back to develop, I have some ideas:

- Improve the templates to generate more technologies, as VS 2008 projects, solutions, templates for Struts 1/2, Spring, better mapping for Hibernate, NHibernate, and code generation for the new ASP.NET MVC.

- Write posts about these templates

- Publish the new examples

- Write down a useful documentation (now, it’s minimal: you can found more details in this posts)

Enjoy the project!

Angel “Java” Lopez
http://www.ajlopez.com/en

Daniel Vaughan has published a very interesting project (LGPL license) at Codeproject:

Legion: Build your own virtual super computer with Silverlight

Daniel is one of the 40 CodeProject MVPs for 2008. He is a software developer based in Canberra Australia, and Prague in the Czech Republic. (Thanks to Arvindra Sehmi, that sends me the article’s link).

It’s a project that uses Silverlight, the new Microsoft technology that runs in the browser, exposing .NET framework to JavaScript and other languages (in the 2.0 version).

According to the CodeProject article:

Legion is a grid computing framework that uses the Silverlight CLR to execute user definable tasks. It provides grid-wide thread-safe operations for web clients. Client performance metrics, such as bandwidth and processor speed, may be used to tailor jobs. Also includes a WPF Manager application.

Recently, I posted about Agents in a grid. Legion puts the agent code inside the browser, using Silverlight as a host environment. The server implements a JsonGridService that is accesible via web services from client side. It serializes the results using JSON (JavaScript Object Notation). One example method from that web services (JsonGridService.asmx.cs):

        [WebMethod]
        [ScriptMethod(ResponseFormat = ResponseFormat.Json)]
        public TaskDescriptor StartNewJob(Agent agent)
        {
            try
            {
                TaskDescriptor descriptor = GridManager.GetDescriptor(agent);
                return descriptor;
            }
            catch (Exception ex)
            {
                HandleException(ex.Message, agent, ex);
            }
            return null;
        }

An instance of TaskDescriptor has a Type and a Job. The Type is a string describing the full name of the class and assembly to load and run in the client. The compiled assembly must reside in the ClientBin directory in the web server application. Job instance is a bit more complex: it’s a message, containing an arbitrary object to process in the client.

The code to run in the browser must inherits from SlaveTask. The client obtains a TaskDescriptor, creates an object according to the Type in TaskDescriptor, and runs it:

static void LoadAndRunTask()
{
    TaskDescriptor info = gridService.StartNewJob(CreateAgent());
    if (!info.Enabled)
    {
        return;
    }

    log.Debug("LoadAndRunTask() for Job Id " + info.Job.Id);

    Type type = Type.GetType(info.TypeName, true, true);
    if (slaveTask != null)
    {    /* Detach last task. */
        slaveTask.Complete -= taskBase_Complete;
        slaveTask.ProgressChanged -= taskBase_ProgressChanged;
    }
    slaveTask = (SlaveTask)Activator.CreateInstance(type);
    slaveTask.Initialise(info);
    OnTaskChanged(EventArgs.Empty);
    slaveTask.Complete += taskBase_Complete;
    slaveTask.ProgressChanged += taskBase_ProgressChanged;

    Thread newThread= new Thread(
        delegate
        {
            slaveTask.RunInternal();
        });
    newThread.Priority = ThreadPriority.Lowest;
    newThread.Start();
//    ThreadPool.QueueUserWorkItem(
//        delegate
//            {
//                slaveTask.RunInternal();
//            }
//        );
    progressTimer.Enabled = true;
}

 Note the use of Type.GetType to load the type. Silverligth uses the ClientBin directory on server side as one of the source for assembly loading. That’s the trick. Activator.CreateInstance creates the instance of the SlaveTask, and a new thread is launched to run the RunInternal method on that instance (for some reason, Vaughan commented the code that used the ThreadPool; I guess that he prefers to manage the priority in code).

In the article, more artifacts are described, as a WPF application, the Legion Manager, that allows monitoring of the grid. But now, let’s examine some concepts and alternatives.

Gridifyng the browser

The great idea, from Vaughan, is to use the browser as a host application for grid node tasks. This idea allows the use of any machine as a grid node (altought Vaughan’s approach needs the support of Silverlight). In the following paragraph, we’ll back to basic, to analyze the full picture implied in this kind of solution.

In a grid computing application, we must resolve the following problems (I’m simplifying the scenario landscape: we could have inter node communications too):

- How to program the task to run (languages, technologies…)

One of the options to the first problem, is to use a specialized language, dedicated to grid computing. Another alternative is to use a main stream language and technology, like Java or .NET. An another one: use scripting/dynamic languages, now supported in Java 6 and in .NET DLR.

- How to inject code to the grid node

If the host node application is a browser, we can program in Java (applets reloaded!), or .NET (now with Silverlight), or even in Flash. To inject the code, the technology at browser can remotely load .jars or compiled assemblies, or it can receive string with source code in dynamic languages, and run it as is.

- How to send task data to the grid node

Well, the data can be serialized, JSON is one method, XML is another one. But, now, another question emerges: who is in charge in sending the data? The server can send the data, but this implies that the client has some listening method. Or the client can poll the server, asking for new task data to process. I don’t know if an Java applet can use a ServerSocket, or if a Silverlight code can open a listening socket. One option to explore, is to have a WCF duplex channel in Silverlight client. Today, the sure option is: the client poll the server. That is the way Legion works.

- How to send back result from the node to a server

This question is an easier one. The data is send using JSON, XML, any serialization technology, to a web service in the server.

Conclusion

Each year, the browser is a more powerful application. Gridifying the browser, using the browser execution capabilities, it’s a promising idea, that deserves more exploration. Security issues, serialization, and the decision about push/pull model on message, are the points to research in more detail.

The next time you open the browser, ask its name. It could be “Legion”….

Angel “Java” Lopez
http://www.ajlopez.com/en
http://www.msmvps.com/lopez

These days, I wrote new examples for my open source project AjGenesis

http://www.ajlopez.com/ajgenesis
http://www.codeplex.com/ajgenesis

Using AjGenesis, we can produce, starting from our own model, applications for ASP.NET 1.x/2.x, JSP, using SQL Server in .NET, with ADO.NET or NHibernate, or MySql with Java and Hibernate. The sample generated applications use a layered architecture (but remember: these are sample applications, you can write and use your own models, tasks, and templates). Other option: it can generate layers following ideas from Eric Evans’ Domain-Driven Design.

I’m still writing those examples, but now, you can download the current version from

AjGenesisExamples3New.zip

The examples require the use of AjGenesis last release:

AjGenesis Version 0.4.3

but you can try the current version under development:

AjGenesis Version 0.5

In this post, I want to comment how AjGenesis works and some of its inner structure, using one of those examples. Although I will concentrate in producing VB.NET code, you’ll find that the example is able to produce C Sharp and Java/JSP solutions, as well.

First, let’s review first some points about the project. The big picture:

AjGenesis Version 0.4.3 is written using VB.NET V1.x, but version 0.5 is the current version under development, and it was rebuilt using .NET 2.0. The project is open source, and has a BSD-like license, that allows to use it in any project what you want: you only need to comply with the license (in short, put some reference to the original project). It is possible to be used like a library, invoked from your own project, or it is possible to be invoked from command line, or even it can be used from using NAnt (this tool gives you a better organization of the code generation tasks).

There are several samples in the page of the project (inside the project zip source file and in additional files). They generate code from a model, invoking tasks and processing templates. The samples generate code PHP, Java, JSP, VB.NET, C#, ASP.NET, and even scripts of procedure and data base stored. I want to emphasize two points:

- The model to use is totally definible by you. It’s a free model, it’s not a fixed one. You can model what you want to model.

- The tasks and templates to apply are totally programmable and controlable. You are in charge. 

That’s make a difference from other generators. We can create our own model, and its own templates and tasks, to generate any text artifacts. Other systems start from the data base, and they only generate a group of predefined text files (as POJOs, plain old java objects, or DAOs, Data Objects Access). But with AjGenesis you can generate any text file you need.

In order to better understandind the free model concept, read a previous post:

Generating Code Hello World with AjGenesis

In that post, the initial steps are described, using a free model, totally oriented to the domain to represent: a typical Hello World application, implemented in different technologies.

Creating an application

Let’s do something more complete in this post. Suppose we need to create a simple solution simple, with two tables, in a MS SQL Server databse, source code in VB.NET 2,0, web interface, layer of services, layer of data, business components and business entities a la Microsoft. We want to generate the solution, the projects, scripts for database creation, and stored procedures. This example is included in the examples AjGenesisExamples3New.zip. First step: to write the model.

The project

In a directory of projects of the examples that accompany this article, there is a Projects/AjFirstExample directory.

In that directory it is the Project.xml file that contains the model.

<Project>
    <Name>AjFirstExample</Name>
    <Description>First Example
    using AjGenesis</Description>
    <Prefix>AjFE</Prefix>
    <Domain>com.ajlopez</Domain>
    <CompanyName>ajlopez</CompanyName>
    <Model>
        <Entities>
            <Entity
            Source="Entities/Customer.xml"/>
            <Entity
            Source="Entities/Supplier.xml"/>
        </Entities>
    </Model>
</Project>

Remember: the model is free. Here we define the templates that we are going to use. The model contains two simple entities: customers and suppliers.

The entities

The XML file is not terribly long: AjGenesis allows that any node of the model is specified apart in a file. This is a criterion that I have used to define how the model is written: the resulting XML does not have to hurt at sight, must be understandable and abarcable in a reading.

In the Project.xml, that feature is used in the case of the entities, with the Source attribute. Let us examine an entity, written in Entities/Customer.xml:

<Entity>
    <Name>Customer</Name>
    <Description>Customer Entity</Description>
    <SetName>Customers</SetName>
    <Descriptor>Customer</Descriptor>
    <SetDescriptor>Customers</SetDescriptor>
    <SqlTable>customers</SqlTable> 

    <Properties> 

        <Property>
            <Name>Id</Name>
            <Type>Id</Type>
        </Property> 

        <Property>
            <Name>Name</Name>
            <Type>Text</Type>
            <SqlType>varchar(200)</SqlType>
        </Property> 

        <Property>
            <Name>Address</Name>
            <Type>Text</Type>
            <SqlType>text</SqlType>
        </Property> 

        <Property>
            <Name>Notes</Name>
            <Type>Text</Type>
            <SqlType>text</SqlType>
        </Property> 

    </Properties>
</Entity> 

There are attributes of the entities, like its name and description, in plural and singular. This data serve to name them in the resulting pages, or within the code. The properties are the fields to maintain in each entity..

Aside from the entities, in another directory, Technologies, specifies the dependent model of the technology, like VbNet2:

The templates

The templates to use are at the Templates/VbNet2 directory:

They are the templates for generation of code VB.Net 2.0. Also we will find groups for C# 1/2, Vb.NET 1,2, Java (although I’m thinking to drop templates for .NET 1.x: I don’t see any reason to maintain these). There are templates to use Nhibernate, Hibernate, JSP, MySql, and concepts of Domain-Driven Design, too. Let’s review a template as an example, the one that generates the organization in Visual BASIC, EntityVb.tpl:

<# 

message "Generating Entity ${Entity.Name}" 

include    "Templates/VbNet2/VbFunctions.tpl" 

include    "Templates/VbNet2/Prologue.tpl"
#> 

'
'    Project ${Project.Name}
'        ${Project.Description}
'    Entity    ${Entity.Name}
'        ${Entity.Description}
'    
'

Public Class ${Entity.Name} 

'    Private Fields

<#
for each Property in Entity.Properties
    message    "Procesando Campo ${Property.Name}"
#>
    Private m${Property.Name} as ${VbType(Property)}
<#
end for
#> 

'    Default Constructor

    Public Sub New()
    End Sub 

'    Public Properties

<#
for each Property in Entity.Properties
    message    "Procesando Propiedad ${Property.Name}"
#>
    Public Property ${Property.Name}() as ${VbType(Property)}
        Get
            Return m${Property.Name}
        End Get
        Set(ByVal Value As ${VbType(Property)})
            m${Property.Name} = Value
        End Set
    End Property
<#
end for
#> 

End Class 

Like before, control structures are used. XML is the serialized format of the model. During the code generation process, the model is loaded in memory, ready to be accesible via dynamic variables.

The steps

We have more files to generate: from the pages ASPX, and their associated code, the projects of facade on watch, organizations, access to data, the file of solution, and more. In order to automate this generation, the example has several files of tasks, in the Tasks directory, where the steps are described to execute. There are two great tasks: the steps to execute independently of the chosen technology, like completing the model, reviewing it, and the employees of the technology, like generating such file JSP or ASPX, depending on if we want Java or .NET.

The task of completing the model is in charge of Tasks\BuildProject.ajg, that begins with:

'
' Build Project
'    Complete the Project Data
'    Project must be loaded in global variable Project
'

PrintLine "Completing Project ${Project.Name}" 

include "Templates/EntityFunctions.tpl"
include "Templates/Utilities.tpl" 

if not Project.Title then
    Project.Title = Project.Name
end if 

if not Project.Version then
    Project.Version = "1.0.*"
end if 

if not Project.SystemName then
    Project.SystemName = Project.Name
end if 

The template includes some auxiliary functions, and then, it begins to complete the model that resides in the Project variable. For example: if the project lacks Project.Title the program set Project.Name as Project.Title. The program continues:

for each Entity in Project.Model.Entities
    PrintLine "Entity " + Entity.Name 

    for each Property in Entity.Properties
        PrintLine "Property " & Property.Name
        if Property.Type="Id" and not Property.SqlType then
            Property.SqlType="int"
        end if
        if Property.SqlType and not Property.SqlColumn then
            Property.SqlColumn = Property.Name
        end if
        if Property.Type="Id" and not Entity.IdProperty then
            Entity.IdProperty = Property
        end if
        if Property.Reference then
...

After that process, the technology tasks are executed. The example use Tasks\BuildVbNet2.ajg, here a fragment:

<# 

include "Templates/Utilities.tpl"
include "Templates/VbNet2/UtilitiesVb.tpl" 

message "Creating Directories..." 

FileManager.CreateDirectory(Project.BuildDir)
FileManager.CreateDirectory("${Project.BuildDir}/Sql")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.Entities")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.Entities/My Project")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.Data")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.Data/My Project")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.Services")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.Services/My Project")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.Business")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.Business/My Project")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.WebClient")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.WebClient/App_Themes")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.WebClient/App_Themes/Default")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.WebClient/Admin")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.WebClient/Controls")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.WebClient/MasterPages")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.WebServices")
FileManager.CreateDirectory("${Project.BuildDir}/Src/${Project.Name}.RemoteServices") 

message "Defining Solution and Projects..." 

...

In this fragment, the directories necessary are created to lodge the solution. The name of the directory is extracted of the model from Project.BuildDir. Note that ${ } in a string is used to expand the inside expression into the string value.

Technology Model

Under the directory Project\AjFirstExample\Technologies they are some models that describes the technology parameters to use:

Let’s examine VbNet2.xml:

<Technology>
    <Programming>
        <Dialect>VbNet2</Dialect>
    </Programming>
    <Database>
        <Dialect>MsSql</Dialect>
        <Name>AjFirstExample</Name>
        <Username>sa</Username>
        <Prefix>ajfe_</Prefix>
        <Host>(local)</Host>
    </Database>
</Technology>

These data is used during the template generation phase. It indicates the language, and the database to use.

Generating the solution

We could send the tasks from the command line, but we have a .build file for Nant, one for each technology to generate. We execute the tasks build, buildsql, and deploysql of AjFirstExampleVbNet2.build with the command line:

nant -buildfile:AjFirstExampleVbNet2.build build buildsql

(You must adjust the line

<property name=”ajgenesis.dir” value=”…./AjGenesis-0.5″/>

to reflect your AjGenesis installation directory)

Note the following lines in the build file:

    <target name="loadtasks" description="loads AjGenesis tasks"