Hi all, Repository[^] pattern with Unit of Work[^] and Dependency Injection, seem to be a universaly accepted solution for ASP.NET + EF applications. There are still some minor questions that I would like to ask here. Consider the following method from a generic IRepository interface:

public IQueryable Get() where T : class
{
return Set();
}

I wonder, is it a good practice in reveling IQueryable interface to the callers. This means that the calling method can add filters/paggin/sorting, but at the same time the repository itself becomes not unit testable (you cant rely anymore on what the repository will actually return). The caller can even add a non-translatable predicate that will break at runtime. So, is it a good practice in hiding IQueryable interface from the caller, and defining IEnumerable interface instead? Also, I wonder how do you perform integration tests on repositories? Do you use SQL CE4, full SQL Server, or just Fake objects on DbContext? Appreciate your response, With regards

Good point.

Cool stuff, like the sourceObject solution. Though, just source sounds even better.

Thanks for suggestion, I was always tempted to append "to". In that case it would be toEncrypt. But it scratches my ear badly. I really like encrypt but its a verb, and it breaks the convention a little. As for your CommandDoSomething.DoSomething example. I might have misunderstood you, but I prefer not appending action name to class name or viceversa:

//prefer
User client = User.Create();
//instead of
User client = User.CreateUser();

Anyway, I believe its a matter of preference :) Regards

Hi guys, The question is of pure interest: how do you prefer naming a variable/parameter of type Object? Most of the time it depends upon the logic of its usage, but not always. Sometimes you just want to express a generic entity:

public byte[] Encrypt(Object x)
public byte[] Serialize(Object o)

Its tempting to name x as obj or objToEncrypt, but it sounds more like a Hungarian notation. So how would you consider naming it?

Congratulations Nick!

Congrats Nish!

At your age I would rather look for knowledge than money. Anyway, you might try finding a job at www.odesk.com, or try your skills at any programming competitions (Google Code Jam, Top Coder, etc.).

Good remark.

Yeap, its like

private string resign(){
//return "Burn in Hell!"
return "Best wishes!"
}

:)

Everybody that leaves our office sends a corporate email to all employees, with a last "thank you" :)

You need to create an inter-process memory map file with MemoryMappedFile.CreateNew method. In this way you'll be able to access it from different processes.

This one would be even more subtle :cool:

if ( !( !( !( !(a==null) == true ) == false ) == true)) {...}

Related to the above: flag = bool(int(flag) - 1);

How about this piece of code: if (a == null == false) { } Spent some time before figuring out it is a version of if (а != null) {}

Check this. It might be helpful http://urenjoy.blogspot.com/2009/08/database-installation-vs-setup-custom.html[^]

Try to set AutoScaleMode of the form to NONE. Not sure if it helps, but there is a good article related to scaling http://en.csharp-online.net/Layout[^]

All in all, you have Delete, Insert, Update in a class, so you can declare an Event (Modified), and subscribe to it. When it fires, call setRecords sub. In C# it looks as follows:

public event EventHandler DatabaseDataModified;

protected virtual void OnDatabaseDataModified(EventArgs args)
{
if(DatabaseDataModified != null)
DatabaseDataModified.Invoke(this, args);
}

Call OnDatabaseDataModified method everytime a record is updated in the database. Subscribe to DatabaseDataModified event and call setRecords sub, once the event handler fires. For a more detailed review check publish/subscribe model.