Re: Architectural question (crisis) - C# Discussion Boards

29-Apr-04 21:48

Hi everyone, i'm having a bit of an architectural crisis.

I want to start an application that is layered. I cannot use datasets, only an object model. I want the following layers: DAL, Common, Services, facade(not nessessary), and web/webservices.

Where do i put the object model (objects like customer, questionlists, employees,...). In the common layer so i can get filled employee's out of the DAL or do i put the OM in the services layer and fill them there?

The first has the advantage that i only need my update delete and insert take 1 parameter (employee) otherwise with the OM in the services layer i need my DAL funtions to take all the parameters.

What do you do?

Thank you
Jan Vercauteren.

OVERVIEW

-----------------------------|---------------|
| DAL | |
-----------------------------| |
| SERVICES | COMMON |
|----------------------------| |
| FACADE | |
|----------------------------| |
| GUI (web) | |
|----------------------------|---------------|

www.agilis.be

Anonymous29-Apr-04 23:05

29-Apr-04 23:05

Hmmm...4 tier architecture? What is facade layer, anyways?

Jan Vercauteren wrote:
want to start an application that is layered. I cannot use datasets, only an object model. I want the following layers: DAL, Common, Services, facade(not nessessary), and web/webservices

What do you mean by common layer??

Well, you could use the collection objects in your services. You can combine objects and services in the service layer.

WEB -- Front End

WEB -- FACADE

WEB -- SERVICES | COMMON

WEB -- Front End

Anonymous29-Apr-04 23:06

29-Apr-04 23:06

Jan Vercauteren30-Apr-04 0:37

30-Apr-04 0:37

Facade layer is a sort of "grouping" layer.

Lets say you have a customer and a orderEmployee.
A customer can add an order
a orderEmpoyee can add an order and approve orders.

You have an order object wich has the 2 functions + a couple of internal ones.

You create a customer facade with only te first operation in it and a orderEmployee facade with both in it.

This makes an application easier to secure and to understand for other people.

This looks not nessessary but if you have 10+ roles and 50 objects and then times the amount of functions witch are not all to be exposed to the client...

jan

www.agilis.be

bjoernen30-Apr-04 1:54

bjoernen

30-Apr-04 1:54

Hello Jan.

The major reason for a DAL is data abstraction; it should deliver data to your application in a data source-independant way. Your OM contains the data transport classes that the DAL uses to deliver data to the service layer, so the OM should rest inside the DAL. The DAL should also contain factory/accessor classes that connects to data sources, obtains data, instantiates OM classes and fills them with the data and delivers them to the service layer.

The "Common" is not needed.

Regards,
Björn Morén
Stockholm, Sweden

Jan Vercauteren30-Apr-04 2:46

30-Apr-04 2:46

Hi Björn

Thank you for the reply.

I asked this because msdn is screaming black and white about the subject of custom objects.

The say put the creation static in the objects, who then ask the dal to create them. Or put the objects in the DAL like you do. Or use a common layer.

Do you have an example or uml scheme?

Another question: Insert update and delete, are those methods inside the custom object?

Thank you

(to respond the the question above this post: a common layer holds classes that can be used in all layers)

www.agilis.be

Heath Stewart30-Apr-04 4:05

30-Apr-04 4:05

The deal with statics that's been on the MSDN site for a while is for a provider pattern. Basically, you have an abstract class with a static method that reads providers from, say, a .config file and creates instances of those classes (that derive from the abstract base class) and return the instance as an abstract base class, something like the following pseudo-code:

public abstract class DataProviderBase
{
  // Logical instance methods, like those to get and set data, information
  // about the provider, etc.
  public static DataProviderBase GetProvider()
  {
    // Read provider configuration from .config file that must extend
    // this abstract base class and return it.
    return GetProviderFromConfig();
  }
}

There is an article posted recently on MSDN that goes into greater depth: Provider Model Design Pattern and Specification, Part 1[^].

Microsoft MVP, Visual C#
My Articles

Jan Vercauteren30-Apr-04 4:55

30-Apr-04 4:55

I just want to know where to put the object model (the classes like customers, orders, ...) in the n-tier model.

Common , Data or business layer. (on msdn they put datasets is the common.
thanks
Jan

www.agilis.be

Heath Stewart30-Apr-04 5:10

30-Apr-04 5:10

I'd prefer to put the object model you refer to in the DAL and use the provider pattern for those objects to get/set data in a data source (such as a database), effectively splitting up the data layer into two tiers (although that depends on your scope of a tier). The business layer typically defines how those objects interact with each other. Again, depending on your view of what comprises a tier, you might consider that those objects are either in the business layer or the data layer. Also keep in mind that N-tier isn't limited to 3 tiers.

For example, our next major version of our application that is currently being sketched out consists of many layers, such as a presentation layer, a business layer, a data objects layer coupled with a data provider layer, with a server layer to provide remotable data services for our implementation (while other companies can choose to override or replace whatever layer they want). It's a pretty fractured design, but a pretty big N-tier model allowing for maximum (IMO) extensibility.

Microsoft MVP, Visual C#
My Articles

bjoernen30-Apr-04 11:51

bjoernen

30-Apr-04 11:51

I'm sure there are many ways to solve this, and much has to do with taste.

For most applications I prefer a 3-tier solution: data, business and facade/presentation, but that is of course depending a lot on the scenario.

The data tier consists of two kinds of classes, none of them static:

1. The objects in the object model (usually maps to tables in a relational database). Those are very much like structures, with no logic in them, just data fields for transport. For example the Customer class. Take a look at xsd.exe to generate those out of schemas.

2. Factory/accessor classes that creates and delivers object model instances, or recieves object model instances. For example CustomerFactory (or CustomerDB if you don't like the "concrete factory" pattern analogy). Those classes contain logic for retrieving data and updating data. If the data resides in SQL server they handle connection objects and have all the insert, update and select in them (or SP's handle that). If the data comes from a web service, they contain web service consumer proxies etc.

Example of methods:
public Customer FindByCustomerID(int id)
public OperationResult Save(Customer[] customers)

I think the most important design decision for you to consider is that the data-tier should not be aware of the other tiers, and it shall reveal nothing about the actual data sources to the other tiers. It delivers the data to whoever needs it, period. Much more can be said about the details, and many books have been written on the subject. Good luck with your project.

Regards,
Björn Morén
Stockholm, Sweden

Jan Vercauteren30-Apr-04 13:07

Stupid string to int question

30-Apr-04 13:07

Thanks Björn

I had just read the "mastering XDE book from sybex" It had an overview of different patterns but i just couldn't fit the puzzle together.

Now it's coming clear to me.

If you ever need a favor, just yell

Greets from Belgium
Jan

www.agilis.be

Anonymous29-Apr-04 20:53

Re: Stupid string to int question

29-Apr-04 20:53

using System;

namespace MBoard
{
///

/// Summary description for Class1.
///

public class Class1
{
public Class1()
{
string t;
int p;
t="11";
p= (int)t+1;

System.Diagnostics.Debug.WriteLine(p);
}

static void Main()
{
Class1 c=new Class1();

}
}
}

OK!! here i want to result as 12, not 111.

How could i get it?

Anonymous29-Apr-04 21:05

Re: Stupid string to int question

29-Apr-04 21:05

Never mind!

Syntax

using System;

namespace MBoard
{
///

/// Summary description for Class1.
///

public class Class1
{
public Class1()
{
string t="123";
int p=1;

System.Diagnostics.Debug.WriteLine(int.Parse(t) + p);
}

static void Main()
{
Class1 c=new Class1();

}
}

}

Heath Stewart30-Apr-04 3:29

Re: Stupid string to int question

30-Apr-04 3:29

You should put int.Parse, Convert.ToInt32, or whatever you use in a try/catch block. Never trust user input! Always validate user input and expect the unexpected, even in throw-away utility projects (practice makes perfect!). Not validating input is the most common reason for all the viruses and worms.

Microsoft MVP, Visual C#
My Articles

medcon29-Apr-04 22:41

medcon

29-Apr-04 22:41

Try int.Parce("11") + 1

Crystal Report fields

DougW4829-Apr-04 20:01

DougW48

29-Apr-04 20:01

Sorry to post a Crystal Report question in here, but I don't know where else to ask it, since I'm using it in C#. I have no problem passing datasets to my report...everything works fine. BUT...I can't seem to find a way of passing a simple variable into my report. I just want to print 2 separate dates on my report, that definitely doesn't require an entire dataset just for 2 dates...but I don't know how to put them in my report. The closest I've come is using a parameter field...but then I have to input the dates in that dialog box that pops up, and I don't like that. I just want to do it programmatically so the user never knows how it happened. Any ideas? Thanks

Re: Crystal Report fields

..Hubert..29-Apr-04 22:55

..Hubert..

29-Apr-04 22:55

parameter fields is exactly what you need.
Here is some howto (cut from visual studio help):

Parameter Fields Runtime Customization
You can support user input with parameters in Crystal reports. Such parameters can be used for a variety of purposes. For example:
Base a parameter on a database field and allow the user to specify values for that field to filter data in the report.
Apply conditional formatting to a report by using parameter fields.
Define sort order by using parameter fields.
The following example demonstrates how parameter field values can be set, through code, at runtime. It explains how to set two different parameters. The first is a multiple-value discrete parameter, and the second is a range value parameter.
To Modify Parameter Fields at Runtime
[C#]

// Declare variables needed to pass the parameters // to the viewer control. ParameterFields paramFields = new ParameterFields (); ParameterField paramField = new ParameterField (); ParameterDiscreteValue discreteVal = new ParameterDiscreteValue (); ParameterRangeValue rangeVal = new ParameterRangeValue (); // The first parameter is a discrete parameter with multiple values. // Set the name of the parameter field, this must match a // parameter in the report. paramField.ParameterFieldName = "Customer Name"; // Set the first discrete value and pass it to the parameter. discreteVal.Value = "AIC Childrens"; paramField.CurrentValues.Add (discreteVal); // Set the second discrete value and pass it to the parameter. // The discreteVal variable is set to new so the previous settings // will not be overwritten. discreteVal = new ParameterDiscreteValue (); discreteVal.Value = "Aruba Sport"; paramField.CurrentValues.Add (discreteVal); // Add the parameter to the parameter fields collection. paramFields.Add (paramField); // The second parameter is a range value. The paramField variable // is set to new so the previous settings will not be overwritten. paramField = new ParameterField (); // Set the name of the parameter field, this must match a // parameter in the report. paramField.ParameterFieldName = "Customer ID"; // Set the start and end values of the range and pass it to the // parameter. rangeVal.StartValue = 42; rangeVal.EndValue = 72; paramField.CurrentValues.Add (rangeVal); // Add the second parameter to the parameter fields collection. paramFields.Add (paramField); // Set the parameter fields collection into the viewer control. crystalReportViewer1.ParameterFieldInfo = paramFields;

Re: Crystal Report fields

hassan azizi1-May-04 7:51

hassan azizi

1-May-04 7:51

threre is 2 simple way :
1- using textobject for sending value to your report:
create textobject with TextTimeFrom name(or desired name) then
in your code set value for TextTimeFrom textobject
...C# code
((TextObject)ReportOne.ReportDefinition.ReportObjects["TextTimeFrom"]).Text = "12:25";
....

this value is placed on your report because textobject will be placed on report area
if you want to send parameter to conitional approach use formula field and you can place formula field object on your form or not!
second way byy using formula fields
2-
create one formula field with YearValue(or desired name,YearValue just is sample name) name and then set that value in your code like this :
...C# code
ReportOne.DataDefinition.FormulaFields["YearValue"].Text= "value;
(ReportOne is your report ReportDocument object)
....
you see it is very simple just with one line coding
enjoy of it

Web request from Windows Forms user control

Shree29-Apr-04 17:32

Shree

29-Apr-04 17:32

I have written a user control, and displayed it on a web page. There is a server side application that is supposed to provide the user control the data. I can do this through setting a property of the user control. What I am stuck at is how to make the user control request data to the server side application.

A simple example. I have a user control with a label and a timer on it. On each timer tick, I want the user control to request the server side application to populate the label with the current time at the server.

Re: Web request from Windows Forms user control

Heath Stewart30-Apr-04 3:28