What information does LINQPad use to generate Association attributes/Navigation Properties?

Jake_Turner94

I'm connecting to a SQL Server database using LINQPad, which then generates a TypedDataContext.

In an old version of the database, tables (which are not generated through ASP.NET C# code) have navigational properties to other tables.

In the new version a team have developed, there are no navigational properties.

My first instinct was that there were missing constraints, which turned out to be true. After a developer added the foreign key constraints to the table, there were still no navigational properties.

Using ILSpy for LINQPad, I reflected on the type represented in the table, and found that in one object there were properties with Association attributes:

[XmlIgnore]
[Association(Name = "FK_AvailabilityActivity_SFT_WK_Shift", Storage = "_Ds_Shift", ThisKey = "SFT_WK,SFT_WK", OtherKey = "SFT_WK,SFT_WK", IsForeignKey = true)]
public Ds_Shift Ds_Shift
{
    get
    {
        //Error decoding local variables: Signature type sequence must have at least one element.
        return _Ds_Shift.Entity;
    }
    set
    {
        //Error decoding local variables: Signature type sequence must have at least one element.
        _Ds_Shift.Entity = value;
    }
}

And the other, new object does not.

So, my question is:

What information does LINQPad use to generate these properties?

Once I know this, I should be able to add that information and query across tables using navigation properties.

JoeAlbahari

It just reads the foreign key constraints and builds it from that. Sometimes it's unable to infer a name for an association, or the name is duplicated in another association or property, but that's quite rare. Are there a lot of missing associations, or just a few?

Jake_Turner94

Thanks for the speed reply, Joe. Big fan of your book/software.

Every association is missing, but now I know the problem is in the foreign key constraints I know where to focus my investigation.
Now I can compare the information schema constraint table usage for each table involved and get the developer to implement what he needs to.

I doubt there would be any naming/duplicated name issues, it's probably just an oversite when scripting the constraints.

Thanks a lot!