您何时需要在实体框架中包括相关实体？

Question

This seems arbitrary to me when I have to actually .Include() related entities and when I don't. In some cases, EF gives me the info for the related entities without it and in other cases, it can't do anything with the related entities because I didn't include them:

Works without .Include();

This is an example where I'm loading data without .Include();

public class InvoiceService
{
    private ApplicationDbContext db { get; set; }
    public InvoiceService(ApplicationDbContext context)
    {
        db = context;
    }

    public Invoice Get(int id)
    {
        return db.Invoices.SingleOrDefault(x => x.Id == id);
    }
}

public partial class ShowInvoice : System.Web.UI.Page
{
    private InvoiceService invoiceService;

    private readonly ApplicationDbContext context = new ApplicationDbContext();
    protected void Page_Load(object sender, EventArgs e)
    {
        invoiceService = new InvoiceService(context);
        if (!IsPostBack)
        {
            int.TryParse(Request.QueryString["invoiceId"].ToString(), out int invoiceId);
            LoadInvoice(invoiceId);
        }
    }

    private void LoadInvoice(int invoiceId)
    {
        var invoice = invoiceService.Get(invoiceId);
        // Other code irrelevant to the question goes here.
    }
}

Here follows the result which includes the data for the Company associated with the invoice I'm requested:

As you can see, the information for the company definitely comes through but was not explicitly included.

Doesn't work without .Include();

Conversely, I've done some mapping to do with invoices in this same project and I got NullReferenceExceptions when fetching the related entities property values because I didn't .Include() .

This method gets all the approved timesheet entries for the specified company. This viewmodel is exclusively to be used when manipulating the association of timesheet entries for an invoice (so you're invoicing based on the timesheet entries selected).

public List<InvoiceTimesheetViewModel> GetInvoiceTimesheetsByCompanyId(int companyId)
{
    var factory = new TimesheetViewModelsFactory();

    var timesheets = db.Timesheets.Where(x => x.Approved && x.Company.Id == companyId && !x.Deleted).ToList();
    return factory.GetInvoiceTimesheetsViewModel(timesheets);
}

NullReferenceExceptions occurred in the factory that maps the timesheet entities to the viewmodel:

public List<InvoiceTimesheetViewModel> GetInvoiceTimesheetsViewModel(List<Timesheet> timesheets)
{
    var model = new List<InvoiceTimesheetViewModel>();
    foreach (var timesheet in timesheets)
    {
        var start = DateTime.Parse((timesheet.DateAdded + timesheet.StartTime).ToString());
        var finished = DateTime.Parse((timesheet.DateCompleted + timesheet.EndTime).ToString());
        DateTime.TryParse(timesheet.RelevantDate.ToString(), out DateTime relevant);

        model.Add(new InvoiceTimesheetViewModel
        {
            RelevantDate = relevant,
            BillableHours = timesheet.BillableHours,
            Finished = finished,
            Id = timesheet.Id,
            StaffMember = timesheet.StaffMember.UserName, // NRE here.
            Start = start,
            Task = timesheet.Task.Name // NRE here.
        });
    }
    return model;
}

To fix these, I had to change the query that fetches the data to the following:

var timesheets = db.Timesheets.Include(i => i.StaffMember).Include(i => i.Task)
            .Where(x => x.Approved && x.Company.Id == companyId && !x.Deleted).ToList();

Why is Entity Framework sometimes happy to give me data without me explicitly requesting that data and sometimes it requires me to explicitly request the data or else throws an error?

And how am I to know when I need to explicitly include the data I'm looking for and when I don't?

Answer 1

Entity framework uses lazy loading to load child relationships. For lazy loading to work property in the model should be marked with virtual keyword. Ef overrides it and adds lazy loading support.

When you have no virtual property EF has no way to load your child relationship data later, so the only time it's possible to do - during initial data loading using Include .

public class Timesheet
{
    ...
    public virtual StaffMember StaffMember { get; set; }
    public virtual Task Task { get; set; }
    ...
}

Answer 2

It depends on your models. If you have marked relational properties as virtual then you'll need to use .Include so EF knows that you need it. It is Lazy Loading. Preserves machine's memory and DB requests.