Take the 2-minute tour ×
Programmers Stack Exchange is a question and answer site for professional programmers interested in conceptual questions about software development. It's 100% free, no registration required.

As mentioned in this question, I am moving our team towards objects (as opposed to just throwing DataTables and variables around everywhere). I have picked a suitable spot for the project that contains the object definitions, but not sure how I should go about populating some of the objects. In particular, what do I do with classes as properties?

Public Class Class1
    Public Property SomeProperty As String
    Public Property SomeOtherProperty As String
    Public Property SomeKey As Integer
    Public Property Foo As Class2
End Class

In this basic example, I can easily populate the first three properties from a query, but Foo has its own set of properties. Right now, to create a List(Of Class1) I generate the data for all of the Class1 objects, and then use SomeKey to go back and, one object at a time, populate all of the Foo. Certainly this works, but in my case it is rather slow, and it seems that perhaps I am simply doing it wrong.

Am I handling that incorrectly? If so, what should I be doing? The only other thing I can come up with is to create a query (or stored procedure) that returns all of the appropriate data for all of the objects. In my test case that amounts to a list that contains 154 Class1's, with each of them having four properties like Foo, and some of those having classes as properties.

share|improve this question
1  
Edited for clarity. Hopefully my question is answerable now. –  Dave Johnson Mar 10 at 13:30

3 Answers 3

up vote 2 down vote accepted

Your current approach causes what is called as N+1 Select problem (see https://stackoverflow.com/questions/97197/what-is-the-n1-selects-issue).

One way to solve this is to retrieve all required data in the query by using select distinct and left join. This is called eager fetching. In most cases, it is the preferred solution. If you have a deep hierarchy, you may want to create separate finder methods in your repository class such as findStudent() will only populate Student and findStudentForReview() will return Student.subjects.assignments.

Another way to solve this is to create a getter for property Foo that will retrieve data for Class2 if the data hasn't been retrieved before. This is called lazy fetching approach: query for Class2 will be executed only when you need Foo value.

Or, you could use a ORM framework for your platform. You will need extra investment here. The advantage is ORM framework will handle data population and provide you with ready-to-use objects.

share|improve this answer

There is no "best practice" on how to "populate objects", the same way as there is no best practice for how to create /populate a string or integer somewhere in your program. Objects are instances of classes, which can be seen as the equivalence of a user-defined type. You create and use some of them within a small block scope, some of them a function scope, some of them in a class scope, module scope or global scope, just as long as you need the individual objects instance in your program to solve a certain task. Which task that is, and which lifetime for an object is appropriate, depends fully on how the class you created, its responsibilities and the usage scenario.

share|improve this answer

To work with entities, you really should be using either Linq-to-SQL, or Entity Framework. There other ORMs but few are really worth it, they have nothing real to offer over EF6. To do this, you first create a model (or multiple models) of your datasources, and then use that model to query the data source through Linq queries. The ORM acts as an abstraction layer between the data provider and your language of choice. For example, in Linq to SQL, your example object would be queried like so:

' First you instanciate your context. This is the virtual connection to the database. You
' are actually connecting to your data model here, the connection to the database is only 
' made when there is actually a query executing.
Using db As New MyDataModel

    ' Then you define your query. Linq queries work in defered execution. They
    ' are only executed when they are actually used. The type returned is an
    ' IEnumerable(Of Class1)
    Dim c1s = From c1 In db.Class1s
              Where c1.SomeProperty = someValue
              Select c1

    ' The objects retrieved maintain their relational model as well, so accessing
    ' the Foo property is as simple as using the navigation property provided by
    ' the ORM. All of the relationships are taken care for you in the background.
    For Each c1 In c1s
        Console.WriteLine(c1.Foo.SomeProperty)
    Next

    ' Navigation properties are essentially joins, links implemented by the relationships
    ' defined in your object relational model. The following example would be this SQL:
    ' SELECT c2.* FROM Class1s c1 INNER JOIN Class2s c2 ON c1.SomeKey = c2.SomeForeignKey
    Dim c2s = From c1 In c1s Select c1.Foo

End Using

In short you would NEVER populate your object like you were describing, except in cases where you need a flattened output from a query, but even then you would use anonymous types.

' You can also select specific elements from a relationship through anonymous types
' Here we just define the output type as we go. This is the equivalent to selecting
' individual fields in a SQL select.
Dim anonTypeObj = (From c1 In db.Class1s
                   Select New With {.c1SomeProperty = c1.SomeProperty,
                                    .c2SomeOtherProperty = c1.Foo.SomeOtherProperty}).First

Console.WriteLine(anonTypeObj.c1SomeProperty & " : " & anonTypeObj.c2SomeOtherProperty)
share|improve this answer
    
Unfortunately, switching to an ORM is just not feasible. –  Dave Johnson Apr 14 at 13:12
    
Well, you ARE switching from datatables to objects yes? So why isn't Entity Framework an option here? I mean the only thing you have to do is add references to your application, create a model of the data structure you need to use, and EF creates all the classes and helpers you need to access your data. In my opinion, trying to implement it all yourself is a waste of time. –  Drunken Code Monkey Apr 14 at 13:15
    
It's a matter of experience and learning curve. None on the team (myself included) has any experience at all with implementing or using an ORM. While doing it this way is not exactly easy, it is more familiar. –  Dave Johnson Apr 14 at 13:18
    
Then you doom yourself. You will ignore an existing open-source platform with all the features you need, try to implement it all yourself, fail miserably and end up spending more time on your EF clone than on your application. Not knowing something is no excuse for not learning it, especially when your alternative is a home-grown sub-standard version of the same thing. Do you want to design a data application or do you want to design an ORM nobody but you will ever use? –  Drunken Code Monkey Apr 14 at 13:28
    
If you only understood how difficult it was to convince everyone to move from VSS to TFS, and to upgrade from VS2005 (not even to 2012, when we did this last fall), and how much I have to fight to use feature requests as an excuse to develop in VB instead of FoxPro (6!!!!)... We are a company running on 15 year old technology. I take it one fight at a time, and moving to an ORM is not one that I would win. –  Dave Johnson Apr 14 at 13:49

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.