You searched for the word(s): userid:9938

More Search Options

RSS Available

• Re: Best Way to Implement OOP Search and Filtering

  If you follow GRASP "Information Expert" principle, then the UserManager should have the responsibility of filtering and sorting the collection. Because - the UserManager class aggregates (or has the necessary information) the User objects. However you have another problem. Is your collection going to hold all the 80K objects in itself? You need to find a way of reducing the number of objects loaded in the collection and at the same time be able to do an effective filter or sort. The following method

  Posted to Architecture (Forum) by biswajitdash on 6/22/2006

• Re: Inheritance - Weird issues I've never seen

  Its about package access mechanism that .NET has. If you are trying to refer any TYPE in another package (assembly), you MUST refer that from project. Otherwise in compile-time how will the C# compiler know where to look for the definition of base-class? Its not going to search all possible assemblies that you may have on the system. You need to tell it where to look for. Its not specific to .NET. Its there for almost all similar scenarios for years.

  Posted to Architecture (Forum) by biswajitdash on 6/13/2006

• Re: Need help on design pattern

  You need to break your single table into multiple ones. It will almost match the class structure you are designing. Create a common table, that has all the common information. Create provider specific tables, with provider specific information. Link the common table and provider specific tables through PK/FK constraints, for data consistency. This is what rsmoke21 has suggested as one options in the last post. So, the tables will look like below. tbl_MemberInfo -------------- INT i_member_id - IDENTITY

  Posted to Architecture (Forum) by biswajitdash on 6/2/2006

• Re: Class design and N-tier

  Even at the theoretical level when discussing pure OOP, I would argue that .Save in the Student object is impure and incorrect. It may be that OOP books do this, but if they do, it is either incorrect or, more often, for the sake of simplicity. Similarly, OOP books might have a try/catch that just sends a message to the user saying "Error!" and not actual enterprise logging. We agree on concept and practical aspects for sure. Just one correction I would like to make. In my original post I had already

  Posted to Architecture (Forum) by biswajitdash on 5/31/2006

• Re: Class design and N-tier

  I don't get how you differentiate between reusability and OO. They are intertwined. Totally self-sufficient is rarely good in OO, because it means everyone is duplicating code that should be externalized into its own object. What you are saying is the equivalent of saying there's nothing wrong with an employee delivering letters by hand instead of using the post office. Of course there is something wrong... someone is wasting time and writing code that may not even be optimized. Furthermore, it is

  Posted to Architecture (Forum) by biswajitdash on 5/30/2006

• Re: Class design and N-tier

  You are right in your confusion. In an ideal OO class design a class will be self-sufficient. However over a period, there has been lot of architectural changes for better re-use of code or components. Thats what has resulted in a layered design approach and design changes. Suppose you need to design a Student class with pure OO in mind. Then your class will have fieds and methods like - Name, Address, GetStudentDetail(), SaveStudentDetail(), AddNewStudent(), RemoveStudent() etc. If you look closely

  Posted to Architecture (Forum) by biswajitdash on 5/30/2006

• Re: Product made of many products

  To add to my last post. Yes you right in treating KIT as a template. Once all the correct selections are done, you compose the final product. From display point of view you can infact create an overall template composit, for the user to manipulate. The overall template can look like the visual in the previous posts. Only you need to add one property say IsChangeble - that represents if the specific composit/leaf can be modified. Based on that the UI can be rendered. Once the user has done all selections

  Posted to Architecture (Forum) by biswajitdash on 5/30/2006

• Re: Product made of many products

  The question you are asking is more to do with how the user sees the products for selection, than the actual composite. You can take any approach while doing the selections, but the final composition will look like the visual I mentioned in the last post. Thats the actual product you will be selling. How you handle it on the user interface is a different task.

  Posted to Architecture (Forum) by biswajitdash on 5/30/2006

• Re: Advantages of using Interfaces

  Simple OO rule: Classes are inherited. Interfaces are implemented . No need to get confused on who quoted what. Come basck to the basics. Because in case of inheritance you are able to override a method of base class. It INHERITS all concrete implementations available in the base class. However in case of interfaces, the class has to IMPLEMENT all the signatures defined as a contract. Nothing is available for inheritance. And modern OO languages like Java or C# etc., DO NOT SUPPORT MULTIPLE INHERITANCE

  Posted to Architecture (Forum) by biswajitdash on 5/29/2006

• Re: Need help on design pattern

  Is there any way where in Main class, i can create member object WITH OUT using IF or SWITCH. Like using some kind of reflection or Activator.CreateInstance() method. Most of the time it is alright to use if/else or switch/case statements in these scenarios. Even if you are hard-coding some logic, the conditional variance has been effectively moved out into a separate different class/method. To make it more elegent and re-usable, do something like below, instead of putting the code in the Main class

  Posted to Architecture (Forum) by biswajitdash on 5/29/2006