Save partial data (update) with Entity Framework - c#

So I'm trying to do an update to an Entity model but it's not going as well as I'd like. I'm only trying to update a few fields in my model, but when I have the following code run, I get a SqlException complaining that there are fields (that I'm not trying to edit) that do not allow nulls. It's seeming like it's trying to create a new row in the database, instead of just updating the existing one? Not too sure what to make of it. Any help would be appreciated.
[HttpPost]
public ActionResult Budget(Proposal proposal)
{
if (ModelState.IsValid)
{
db.Proposals.Attach(proposal);
db.ObjectStateManager.ChangeObjectState(proposal, EntityState.Modified);
db.SaveChanges();
return RedirectToAction("Details", new {id = proposal.ProposalID});
}
return View(proposal);
}
Here's the SqlException that I'm getting: "System.Data.SqlClient.SqlException: Cannot insert the value NULL into column 'Ulid', table 'casurg2_dev2.dbo.Proposals'; column does not allow nulls. UPDATE fails.
The statement has been terminated."
And here's the view in question: http://pastie.org/7984222

Ok, the source didn't really help as much (and btw - this is SO, you can paste your code here; no need to be pasting it elsewhere ;)
I did ask for the class definition of the Proposal entity aswell... but in it's absence i can only suggest [at a guess] that your put the original value of Ulid as a hidden field alongside your ProposalId like such:
#Html.HiddenFor(model => model.ProposalID)
#Html.HiddenFor(model => model.Ulid)
so that the model binder can wire it up correctly on the post.
There are 2 ways i can think of to get around the effort with all the hiddenfor's
Use the Add Controller Wizard to set the template to EntityFramework - then modify the related View code. throw away what you don't want etc.
Change the Action method to re-fetch the Proposal record (using its ProposalId) and then bind the changed data to it. This is not the kinda advice i should be giving (an extra, unnecessary round trip to the db)
[HttpPost]
public ActionResult Budget(Proposal proposal)
{
if (ModelState.IsValid)
{
var proposalToMerge = db.Proposals.Find(proposal.ProposalId);
UpdateModel(proposalToMerge);
// don't need this anymore as entity instance is being tracked
// db.Proposals.Attach(proposalToMerge );
// db.ObjectStateManager.ChangeObjectState(proposalToMerge , EntityState.Modified);
db.SaveChanges();
return RedirectToAction("Details", new {id = proposal.ProposalID});
}
return View(proposal);
}
Or summing like that. Untested.

Related

Update record if it exists

I'm new to EF, and I'm trying to understand the best way to handle inserting and updating data. Just for some context, I'm using the .net mvc website boiler plate, and I've created a customers table with a 1:1 relationship to aspnetusers. I've created a view to manage customer data.
Here is my HttpPost ActionResult:
[HttpPost]
[ValidateAntiForgeryToken]
public async Task<ActionResult> AddCard(External.Stripe.Customer customer)
{
if (!ModelState.IsValid)
{
return View(customer);
}
External.Stripe.CustomerOperations co = new External.Stripe.CustomerOperations();
customer = co.Create(customer);
var user = UserManager.FindById(User.Identity.GetUserId());
customer.UserId = user.Id;
var context = new External.Stripe.CustomerContext();
context.Customers.Add(customer);
context.SaveChanges();
return RedirectToAction("Index", "Manage");
}
I feel like I'm going down the wrong path, because if I make any updates to the customer model, I actually need to check EF if my key already exists, and if it does, run an update instead. I figured EF would have a native way of saving my model if it needs to be updated. I also don't know if this is even where I would persist my model, or if it is better placed somewhere else in my code.
Short Answer
...if I make any updates to the customer model, I actually need to check EF if my key already exists, and if it does, run an update instead.
AddOrUpdate() does just this. From the docs:
Adds or updates entities by key when SaveChanges is called. Equivalent to an "upsert" operation from database terminology
Example
In other words, change your code to this:
context.Customers.AddOrUpdate(c => c.UserId, customer);
context.SaveChanges();
In the above example, the c => c.UserId is an expression specifying that the UserId should be used when determining whether an Add or Update operation should be performed.
Considerations
AddOrUpdate() matches database tables based on an arbitrary, user supplied key value. In the example, this key is UserId. Be sure to use an appropriate key.
AddOrUpdate() updates all entity values and sets database cells to null for properties that a lack value. Be sure to set all the property values of the object (e.g. customer) before using it.
See also
Update Row if it Exists Else Insert Logic with Entity Framework It has a few answers that talk about four or more different approaches.
I personally like a long winded approach of pulling the data object and using if null to determine if it should be added or updated. However, I do have a table with a lot of columns that gets changed frequently due to new or deprecated business rules. So I use the below
var currentModel = dbContext.Table.First(t => t.Id == _id);
dbContext.Entry(currentModel).CurrentValues.SetValues(newModel);
dbContext.SaveChanges();
I get my newModel from my viewModel where I have a ToModel method that returns the DTO. Now all I have to update is my view and viewModel when a change to the database table is made.

EF dbcontext.Entry in ASP.NET app

I know it can be silly question, but although ...
Im learning ASP.NET MVC 4 and got one question about the dbcontext.Entry method, lets assume we got such method :
public ActionResult Edit(Movie movie)
{
if (ModelState.IsValid)
{
db.Entry(movie).State = EntityState.Modified;
db.SaveChanges();
return RedirectToAction("Index");
}
return View(movie);
}
How the db.Entry(movie) determine which exactly row from the database is this "movie" instance? If all properties of this movie parameter object will be changed, how will it be possible to determine which row this instance should be attached to? I got smth on my head like it will determine by the Id which we cant change but not sure, so better if I ask :)
When EF generates the models it includes information on what column is the primary key in the database. db.Entry(model) binds the model to the context using that primary key and whatever primary key it shares in the database will be overwriten by the data in the model when SaveChanges() is called. If the primary key is 0, or empty you must set the State to EntityState.Added instead of EntityState.Modified.
As another answer said this can fail. The reason that binding can fail is that a context can only track a single distinct entity once (entity being a distinct class and id). If you attempt to do the following you'll get an exception:
var dbModel = db.Movie.Find(movie.ID);
db.Entry(movie.ID) // Doesn't matter what you do, this will always fail
Because the db context is already tracking a Movie with the same ID.
This is one of the larger reasons that you don't want to share db contexts between users, because another user attempting to update the same entry will cause the second one to throw an exception. While the transaction will be ACID, you can end up with edge case race conditions (for lack of a better term) between users.
Every Table that used by EF must have a Primary key, unless it is a collection , so the EF know the row by the primary key that stay the same.
This method has a parameter Movie, but it has nothing to do with database. You must update object that you will get from db like:
public ActionResult Edit(Movie updatedMovie)
{
if (ModelState.IsValid)
{
//db=your entities context
//here you will get instance of record from db that you want to update
Movie movie = db.Movie.Find(updatedMovie.Id);
TryUpdateModel(movie);
ctx.SaveChanges();
return RedirectToAction("Index");
}
return View(movie);
}
Maybe there is more efficient way , but I'm using it like this and it works fine.
If code in your question is a version that is supposed to be a working version, then it wasn't working for me when I have been trying to solve a similar problem so I had to solve it by using code above finally.

EF concurrency - rowversion [duplicate]

Closed. This question needs debugging details. It is not currently accepting answers.
Edit the question to include desired behavior, a specific problem or error, and the shortest code necessary to reproduce the problem. This will help others answer the question.
Closed 4 years ago.
The community reviewed whether to reopen this question 1 year ago and left it closed:
Original close reason(s) were not resolved
Improve this question
I am using Entity Framework to populate a grid control. Sometimes when I make updates I get the following error:
Store update, insert, or delete statement affected an unexpected number of rows (0). Entities may have been modified or deleted since entities were loaded. Refresh ObjectStateManager entries.
I can't figure out how to reproduce this. But it might have something to do with how close together I make the updates. Has anyone seen this or does anyone know what the error message refers to?
Edit: Unfortunately I am no longer at liberty to reproduce the problem I was having here, because I stepped away from this project and don't remember if I eventually found a solution, if another developer fixed it, or if I worked around it. Therefore I cannot accept any answers.
I ran into this and it was caused by the entity's ID (key) field not being set. Thus when the context went to save the data, it could not find an ID = 0. Be sure to place a break point in your update statement and verify that the entity's ID has been set.
From Paul Bellora's comment
I had this exact issue, caused by forgetting to include the hidden ID
input in the .cshtml edit page
That's a side-effect of a feature called optimistic concurrency.
Not 100% sure how to turn it on/off in Entity Framework but basically what it's telling you is that between when you grabbed the data out of the database and when you saved your changes someone else has changed the data (Which meant when you went to save it 0 rows actually got updated). In SQL terms, their update query's where clause contains the original value of every field in the row, and if 0 rows are affected it knows something's gone wrong.
The idea behind it is that you won't end up overwriting a change that your application didn't know has happened - it's basically a little safety measure thrown in by .NET on all your updates.
If it's consistent, odds are it's happening within your own logic (EG: You're actually updating the data yourself in another method in-between the select and the update), but it could be simply a race condition between two applications.
Wow, lots of answers, but I got this error when I did something slightly different that no on else has mentioned.
Long story short, if you create a new object and tell EF that its modified using the EntityState.Modified then it will throw this error as it doesn't yet exist in the database. Here is my code:
MyObject foo = new MyObject()
{
someAttribute = someValue
};
context.Entry(foo).State = EntityState.Modified;
context.SaveChanges();
Yes, this seems daft, but it arose because the method in question used to have foo passed to it having been created earlier on, now it only has someValue passed to it and creates foo itself.
Easy fix, just change EntityState.Modified to EntityState.Added or change that whole line to:
context.MyObject.Add(foo);
I was facing this same scaring error... :) Then I realized that I was forgetting to set a
#Html.HiddenFor(model => model.UserProfile.UserId)
for the primary key of the object being updated! I tend to forget this simple, but very important thingy!
By the way: HiddenFor is for ASP.NET MVC.
Check whether you forgot the "DataKeyNames" attribute in the GridView.
it's a must when modifying data within the GridView
http://msdn.microsoft.com/en-us/library/system.web.ui.webcontrols.gridview.datakeynames.aspx
The issue is caused by either one of two things :-
You tried to update a row with one or more properties are Concurrency Mode: Fixed .. and the Optimistic Concurrency prevented the data from being saved. Ie. some changed the row data between the time you received the server data and when you saved your server data.
You tried to update or delete a row but the row doesn't exist. Another example of someone changing the data (in this case, removing) in between a retrieve then save OR you're flat our trying to update a field which is not an Identity (ie. StoreGeneratedPattern = Computed) and that row doesn't exist.
I got this same error because part of the PK was a datetime column, and the record being inserted used DateTime.Now as the value for that column. Entity framework would insert the value with millisecond precision, and then look for the value it just inserted also with millisecond precision. However SqlServer had rounded the value to second precision, and thus entity framework was unable to find the millisecond precision value.
The solution was to truncate the milliseconds from DateTime.Now before inserting.
I was having same problem and #webtrifusion's answer helped find the solution.
My model was using the Bind(Exclude) attribute on the entity's ID which was causing the value of the entity's ID to be zero on HttpPost.
namespace OrderUp.Models
{
[Bind(Exclude = "OrderID")]
public class Order
{
[ScaffoldColumn(false)]
public int OrderID { get; set; }
[ScaffoldColumn(false)]
public System.DateTime OrderDate { get; set; }
[Required(ErrorMessage = "Name is required")]
public string Username { get; set; }
}
}
I had the same problem, I figure out that was caused by the RowVersion which was null.
Check that your Id and your RowVersion are not null.
for more information refer to this tutorial
http://www.asp.net/mvc/tutorials/getting-started-with-ef-using-mvc/handling-concurrency-with-the-entity-framework-in-an-asp-net-mvc-application
I also came across this error. The problem it turned out was caused by a Trigger on the table I was trying to save to. The Trigger used 'INSTEAD OF INSERT' which means 0 rows ever got inserted to that table, hence the error. Luckily in may case the trigger functionality was incorrect, but I guess it could be a valid operation that should somehow be handled in code. Hope this helps somebody one day.
While editing include the id or primary key of the entity as a hidden field in the view
ie
#Html.HiddenFor(m => m.Id)
that solves the problem.
Also if your model includes non-used item include that too and post that to the controller
I started getting this error after changing from model-first to code-first. I have multiple threads updating a database where some might update the same row. I don't know why I didn't have a problem using model-first, assume that it uses a different concurrency default.
To handle it in one place knowing the conditions under which it might occur, I added the following overload to my DbContext class:
using System.Data.Entity.Core.Objects;
using System.Data.Entity.Infrastructure;
public class MyDbContext: DbContext {
...
public int SaveChanges(bool refreshOnConcurrencyException, RefreshMode refreshMode = RefreshMode.ClientWins) {
try {
return SaveChanges();
}
catch (DbUpdateConcurrencyException ex) {
foreach (DbEntityEntry entry in ex.Entries) {
if (refreshMode == RefreshMode.ClientWins)
entry.OriginalValues.SetValues(entry.GetDatabaseValues());
else
entry.Reload();
}
return SaveChanges();
}
}
}
Then called SaveChanges(true) wherever applicable.
The line [DatabaseGenerated(System.ComponentModel.DataAnnotations.DatabaseGeneratedOption.None)] did the trick in my case:
using System.ComponentModel.DataAnnotations;
using System.ComponentModel.DataAnnotations.Schema;
[Key]
[DatabaseGenerated(DatabaseGeneratedOption.None)]
public int? SomeNumber { get; set; }
You need to explicitly include a BoundField of the primary key. If you don't want the user to see the primary key, you have to hide it via css:
<asp:BoundField DataField="Id_primary_key" ItemStyle-CssClass="hidden"
HeaderStyle-CssClass="hidden" />
Where 'hidden' is a class in css that has it's display set to 'none'.
I came across this issue on a table that was missing a primary key and had a DATETIME(2, 3) column (so the entity's "primary key" was a combination of all the columns)... When performing the insert the timestamp had a more precise time (2018-03-20 08:29:51.8319154) that was truncated to (2018-03-20 08:29:51.832) so the lookup on key fields fails.
Just make sure table and form both have primary key and edmx updated.
i found that any errors during update were usually because of:
- No primary key in Table
- No primary key in Edit view/form (e.g. #Html.HiddenFor(m=>m.Id)
I also had this error. There are some situations where the Entity may not be aware of the actual Database Context you are using or the Model may be different. For this, set: EntityState.Modified; to EntityState.Added;
To do this:
if (ModelState.IsValid)
{
context.Entry(yourModelReference).State = EntityState.Added;
context.SaveChanges();
}
This will ensure the Entity knows youre using or adding the State youre working with. At this point all the correct Model Values need to be set. Careful not to loose any changes that may have been made in the background.
Hope this helps.
#Html.HiddenFor(model => model.RowVersion)
My rowversion was null, so had to add this to the view
which solved my issue
public void Save(object entity)
{
using (var transaction = Connection.BeginTransaction())
{
try
{
SaveChanges();
transaction.Commit();
}
catch (OptimisticConcurrencyException)
{
if (ObjectStateManager.GetObjectStateEntry(entity).State == EntityState.Deleted || ObjectStateManager.GetObjectStateEntry(entity).State == EntityState.Modified)
this.Refresh(RefreshMode.StoreWins, entity);
else if (ObjectStateManager.GetObjectStateEntry(entity).State == EntityState.Added)
Detach(entity);
AcceptAllChanges();
transaction.Commit();
}
}
}
I had the same problem.
In my case I was trying to update the primary key, which is not allowed.
I got this error sporadically when using an async method. Has not happened since I switched to a synchronous method.
Errors sporadically:
[Authorize(Roles = "Admin")]
[HttpDelete]
[Route("file/{id}/{customerId}/")]
public async Task<IHttpActionResult> Delete(int id, int customerId)
{
var file = new Models.File() { Id = id, CustomerId = customerId };
db.Files.Attach(file);
db.Files.Remove(file);
await db.SaveChangesAsync();
return Ok();
}
Works all the time:
[Authorize(Roles = "Admin")]
[HttpDelete]
[Route("file/{id}/{customerId}/")]
public IHttpActionResult Delete(int id, int customerId)
{
var file = new Models.File() { Id = id, CustomerId = customerId };
db.Files.Attach(file);
db.Files.Remove(file);
db.SaveChanges();
return Ok();
}
I got that error when I was deleting some rows in the DB (in the loop), and the adding the new ones in the same table.
The solutions for me was, to dynamicaly create a new context in each loop iteration
This will also happen if you are trying to insert into a unique constraint situation, ie if you can only have one type of address per employer and you try to insert a second of that same type with the same employer, you will get the same problem.
OR
This could also happen if all of the object properties that were assigned to, they were assigned with the same values as they had before.
using(var db = new MyContext())
{
var address = db.Addresses.FirstOrDefault(x => x.Id == Id);
address.StreetAddress = StreetAddress; // if you are assigning
address.City = City; // all of the same values
address.State = State; // as they are
address.ZipCode = ZipCode; // in the database
db.SaveChanges(); // Then this will throw that exception
}
I got this exception when attaching an object that didn't exist in the database. I had assumed the object was loaded from a separate context, but if it was the user's first time visiting the site, the object was created from scratch. We have auto-incrementing primary keys, so I could replace
context.Users.Attach(orderer);
with
if (orderer.Id > 0) {
context.Users.Attach(orderer);
}
Well i have this same issue. But this was due to my own mistake. Actually i was saving an object instead of adding it. So this was the conflict.
One way to debug this problem in an Sql Server environment is to use the Sql Profiler included with your copy of SqlServer, or if using the Express version get a copy of Express Profiler for free off from CodePlex by the following the link below:
Express Profiler
By using Sql Profiler you can get access to whatever is being sent by EF to the DB. In my case this amounted to:
exec sp_executesql N'UPDATE [dbo].[Category]
SET [ParentID] = #0, [1048] = NULL, [1033] = #1, [MemberID] = #2, [AddedOn] = #3
WHERE ([CategoryID] = #4)
',N'#0 uniqueidentifier,#1 nvarchar(50),#2 uniqueidentifier,#3 datetime2(7),#4 uniqueidentifier',
#0='E060F2CA-433A-46A7-86BD-80CD165F5023',#1=N'I-Like-Noodles-Do-You',#2='EEDF2C83-2123-4B1C-BF8D-BE2D2FA26D09',
#3='2014-01-29 15:30:27.0435565',#4='3410FD1E-1C76-4D71-B08E-73849838F778'
go
I copy pasted this into a query window in Sql Server and executed it. Sure enough, although it ran, 0 records were affected by this query hence the error being returned by EF.
In my case the problem was caused by the CategoryID.
There was no CategoryID identified by the ID EF sent to the database hence 0 records being affected.
This was not EF's fault though but rather a buggy null coalescing "??" statement up in a View Controller that was sending nonsense down to data tier.
None of the above answers quite covered my situation and the solution to it.
Code where the error was thrown in MVC5 controller:
if (ModelState.IsValid)
{
db.Entry(object).State = EntityState.Modified;
db.SaveChanges(); // line that threw exception
return RedirectToAction("Index");
}
I received this exception when I was saving an object off an Edit view. The reason it threw it was because when I went back to save it, I had modified the properties that formed the primary key on the object. Thus, setting its state to Modified didn't make any sense to EF - it was a new entry, not a previously saved one.
You can solve this by either A) modifying the save call to Add the object, or B) just don't change the primary key on edit. I did B).
When the accepted answer said "it won't end up overwriting a change that your application didn't know has happened", I was skeptic because my object was newly created. But then it turns out, there was an INSTEAD OF UPDATE, INSERT- TRIGGER attached to the table which was updating a calculated column of the same table.
Once I change this to AFTER INSERT, UPDATE, it was working fine.
This happened to me due to a mismatch between datetime and datetime2. Strangely, it worked fine prior to a tester discovering the issue. My Code First model included a DateTime as part of the primary key:
[Key, Column(Order = 2)]
public DateTime PurchasedDate { get; set; } = (DateTime)SqlDateTime.MinValue;
The generated column is a datetime column. When calling SaveChanges, EF generated the following SQL:
-- Region Parameters
DECLARE #0 Int = 2
DECLARE #1 Int = 25
DECLARE #2 Int = 141051
DECLARE #3 DateTime2 = '2017-07-27 15:16:09.2630000' --(will not equal a datetime value)
-- EndRegion
UPDATE [dbo].[OrganizationSurvey]
SET [OrganizationSurveyStatusId] = #0
WHERE ((([SurveyID] = #1) AND ([OrganizationID] = #2)) AND ([PurchasedDate] = #3))
Because it was trying to match a datetime column with a datetime2 value, it returned no results. The only solution I could think of was to change the column to a datetime2:
[Key, Column(Order = 2, TypeName = "DateTime2")]
public DateTime PurchasedDate { get; set; } = (DateTime)SqlDateTime.MinValue;
If you are trying to create mapping in your edmx file to a "function Imports", this can result this error. Just clear the fields for insert, update and delete that is located in Mapping Details for a given entity in your edmx, and it should work.
I hope I made it clear.

MVC 4 C# problems with my controller updating database

I have an issue with my method that updates the database with user information.
during HttpGet it uses a view model to ensure the integrity of the data, and in HttpPost, the data is passed back and is there. (ran some checks with breakpoints and everything was holding the correct user data that they should have been)
However, when I run my .Save() method, this information is not stored into the DB.
I've checked to see if its pointing to the correct database by changing the data manually, it comes up in a list view just fine. I'M missing something, just can't figure out what ><
Below is the code there is form data in all of the relevant view models, but it just doesn't save!
[HttpPost]
public ActionResult HouseCreate(CreateHouseViewModel viewModel)
{
if (ModelState.IsValid)
{
var house = new House();
house.address = viewModel.address;
house.postCode = viewModel.postCode;
house.noOfDisputes = viewModel.noOfDisputes;
_db.Save();
return RedirectToAction("House");
}
return View(viewModel);
}
house is an object of my house database, modelled correctly with correct Primary keys in place (double checked this in the database viewer.)
save calls:
void DataSources.Save()
{
SaveChanges();
}
No errors come up, which makes it even worse.
using (var context = new UnicornsContext())
{
var unicorn = new Unicorn { Name = "Franky", PrincessId = 1};
context.Unicorns.Add(unicorn); // your missing line
context.SaveChanges();
}
Source
Your "_db" object, I assume is entity framework.
This will have a collection on it, maybe _db.Houses ? You need to add the object to that.
Since it appears you are using Entity Framework, you'll need to inform Entity Framework that your model has been updated. Inbetween the HttpGet and the HttpPost methods, the model becomes dissassociated from the database context. I think the following code work to re-attach it to the context so you can save the changes:
context.Entry(myEntity).State = EntityState.Modified;
context.MyEntities.Attach(myModel);
context.SaveChanges();
Edit: Forgot to add the re-attach method part, assuming this is an existing record in the DB that your updating.
This I find is a far more reliable means to adding items to the database:
// Note: db is your dbContext, model is your view model
var entity = db.Entry(model); // Required to attach the entity model to the database
// you don't need the "var entity = " but its useful
// when you want to access it for logging
db.Entry(model).State = EntityState.Added; // Required to set the entity as added -
// modified does not work correctly
db.SaveChanges(); // Finally save

ASP.NET MVC with EF 4.1 Navigation properties

After days studying EF to understand (kinda..) how it works, I finally realized that I might have a big problem.
Imagine that I have two entities: Pais and UF. The relationship between them is Pais (0..1) ... (*) UF. A screenshot: http://i.imgur.com/rSOFU.jpg.
Said that, consider that I have a controller called UFController and it has actions for Edit and Create, which are just fine. My views are using the EditorFor helper (or similar ones) for inputs, so when I submit the form the controller will receive a UF object filled with all the data (automatically) with a reference to an almost-empty Pais. My view code (part of it):
#* UF attributes *#
#Html.EditorFor(m => m.Sigla)
#Html.EditorFor(m => m.Descricao)
#Html.EditorFor(m => m.CodigoIBGE)
#Html.EditorFor(m => m.CodigoGIA)
#* Pais primary key ("ID") *#
#Html.EditorFor(m => m.Pais.Codigo) // Pais id
The controller Edit action code:
[HttpPost]
public ActionResult Edit(UF uf)
{
try
{
if (ModelState.IsValid)
{
db.UFs.Attach(uf);
db.ObjectStateManager.ChangeObjectState(uf, EntityState.Modified);
db.SaveChanges();
return this.ClosePage(); // An extension. Just ignore it.
}
}
catch (Exception e)
{
this.ModelState.AddModelError("Model", e.Message.ToString());
}
return View(uf);
}
When I submit the form, this is what the action receives as uf:
{TOTALWeb.UF}
base {System.Data.Objects.DataClasses.EntityObject}: {TOTALWeb.UF}
(...)
CodigoGIA: 0
CodigoIBGE: 0
Descricao: "Foobar 2001"
ID: 936
Pais: {TOTALWeb.Pais}
PaisReference: {System.Data.Objects.DataClasses.EntityReference<TOTALWeb.Pais>}
And uf.Pais:
{TOTALWeb.Pais}
base {System.Data.Objects.DataClasses.EntityObject}: {TOTALWeb.Pais}
Codigo: 0
CodigoBACEN: null
CodigoGIA: null
CodigoIBGE: null
Descricao: null
UF: {System.Data.Objects.DataClasses.EntityCollection<TOTALWeb.UF>}
The original information (the one on the database) is uf.Pais.Codigo == 716. So, right now I'm receiving the updated information. The problem on that the controller is not upading the FK in the database.
I don't want to set the EntityState from uf.Pais to Modified because the entity itself wasn't changed (I didn't changed the information from that entry), but the relationship was.
In other words, what I'm trying to do is change the value of the FK, pointing the uf.Pais to another instance of Pais. Afaik, it's impossible to change the relationship state to Modified (throw an exception), so I'm looking for alternative solutions.
I've read a bunch of topics I've found on Google about this kind of problem, but I still didn't find a simple and elegant solution. The last ones I read here on stackoverflow:
How to work with navigation properties (/foreign keys) in ASP.NET MVC 3 and EF 4.1 code first
Strongly-Typed ASP.NET MVC with ADO.NET Entity Framework
Getting Error 3007 when I add my Entity Model
I asked a question a few days ago about a similar problem ( Entity Framework 4.1 - default EntityState for a FK? ). I didn't understand how EF works that time, so now a bunch of things look clear to me (that's why I'm opening a new question).
For the Create action I've been testing this solution (proposed by Ladislav on my other question), but it generates an additional select (which can be eventually slow for us):
// Here UF.Pais is null
db.UFs.AddObject(uf);
// Create dummy Pais
var pais = new Pais { Id = "Codigo" };
// Make context aware of Pais
db.Pais.Attach(pais); // <- Executing a SELECT on the database, which -can- be slow.
// Now make the relation
uf.Pais = pais;
db.SaveChanges();
I can replicate this for the Edit (I guess), but I don't want that additional SELECT.
So, in resume: I'm trying to use navigation properties to send data to the controller and save them directly in the database using a fast and easy way (without messing too much with the entity - these ones are simple, but we have huge and very complex ones with a lot of FKs!). My question is: there's a solution that doesn't involve executing another query in the database (a simple one)?
Thanks,
Ricardo
PS: sorry for any english mistakes and any confusions.
Update 1: using BennyM's solution (kind of..)
I tested the following code, but it doesn't work. It throws an exception: "An object with the same key already exists in the ObjectStateManager. The ObjectStateManager cannot track multiple objects with the same key." Probably because Pais is already in the context, I guess?
I'm using a Entities (created by EF) class as context. Also, I don't know what is the method Entry, and I don't know where is it. Just for "fun", I tested this:
// Attach the Pais referenced on editedUF, since editedUF has the new Pais ID, not the old one.
Pais existingPais = new Pais { Codigo = editedUF.Pais.Codigo };
db.Paises.Attach(existingPais);
// Attach the edited UF.
db.UFs.Attach(editedUF);
// Set the correct Pais reference (ignoring the current almost-null one).
editedUF.Pais = existingPais;
// Change the object state to modified.
db.ObjectStateManager.ChangeObjectState(editedUF, EntityState.Modified);
// Save changes.
db.SaveChanges();
return this.ClosePage();
The exception is throwed when I try to attach the editedUF to the current context. I'm working with this idea right now, trying to find other solutions. Also, you're right BennyM, attaching the Pais to the context is not generating an additional SELECT. I don't know what happened that time, it really doesn't do anything with the database.
Still, this is a manual solution: I have to do that for each FK. That's what I'm trying to avoid. You see, some programmers, even if you explain 100 times, won't remember to do that with each FK. Eventually that'll come back to me, so I'm trying to avoid anything that can lead into errors (database or code ones) to make sure everyone can work without any stress. :)
I'm answering my own question because I've found a simple solution (at least in my case). My scenario uses a lot of Views for data input (which means that I have a lot of entities). I needed a simple and easy to use solution, so I deleted my entire Entities EDMX file (Ctrl+A, Delete!).
Then I decided to add again Pais and UF entities, but checking the checkbox for exposing the FK attribute. On first I though they can't work together, but they can, but you need to be a little careful on how to use it. They're now linked with navigation properties and the exposed FK.
The reason I couldn't add the FK attribute is because I was doing it manually. Using the "Update model from database" again checking the correct option it worked flawless.
In my edit view, I'm setting the ID of Pais into the FK attribute, not the Pais.Codigo. The reason why I do that is because the FK attribute is a scalar property and then I can detect changes.
This is the current view code for the Pais input (it's not exactly it, but it's similar to this):
#Html.EditorFor(m => m.PaisCodigo)
Btw, PaisCodigo is the FK. Yes, it can get a little confusing with Pais.Codigo, but we didn't decided any naming rules (yet). Any suggestions on this idea would be appreciated.
The final Edit action code is like this (I removed error processing to make it look simple!):
[HttpPost]
public ActionResult Edit(UF editedUF)
{
if (ModelState.IsValid)
{
// Attach the edited UF into the context and change the state to Modified.
db.UFs.Attach(editedUF);
db.ObjectStateManager.ChangeObjectState(editedUF, EntityState.Modified);
// Save changes.
db.SaveChanges();
// Call an extension (it's a redirect action to another page, just ignore it).
return this.ClosePage();
}
}
This is what is received when I submit the form for editedUF:
{TOTALWeb.UF}
base {System.Data.Objects.DataClasses.EntityObject}: {TOTALWeb.UF}
(...)
CodigoGIA: 0
CodigoIBGE: 0
CodigoPais: 0 <-- new Pais ID!
Descricao: "Foobar 2000"
ID: 902
Pais: {TOTALWeb.Pais}
PaisReference: {System.Data.Objects.DataClasses.EntityReference<TOTALWeb.Pais>}
Sigla: "RI"
Usuarios: {System.Data.Objects.DataClasses.EntityCollection<TOTALWeb.Usuario>}
As you can see, CodigoPais is pointing to the new Pais ID.
About the editedUF.Pais navigation property, there's a small detail. Before attaching it into the context, it's null. But, hey, after adding, this is what happens:
{TOTALWeb.Pais}
base {System.Data.Objects.DataClasses.EntityObject}: {TOTALWeb.Pais}
(...)
Codigo: 0
CodigoBACEN: 1058
CodigoGIA: 0
CodigoIBGE: null
Descricao: "Brasil"
UFs: {System.Data.Objects.DataClasses.EntityCollection<TOTALWeb.UF>}
So, it has been filled. The cost for that should be one query, but I couldn't capture it on the monitor.
In other words, just expose the FK, change it using the View and use the navigation property to make the code a little more clear. That's it! :)
Thanks everyone,
Ricardo
PS: I'm using dotConnect for Oracle as a base for the EF 4.1. We don't use SQL Server (at least for now). The "monitor" I said before was devArt's dbMonitor, so I can see all queries sent to the Oracle database. And, again, sorry for any english mistakes!
If you include the foreign keys in your model. So add a PaisId property to the UF entity, you can directly set it and it will update the association.
using (var db = new Context())
{
db.UFs.Attach(editedUF);
editedUF.PaisId = theForeignKey;
db.Entry(editedUF).State = EntityState.Modified;
db.SaveChanges();
}
Also I've tested the approach you already mentioned in your question, and I don't get that extra select when attaching an entity. So this works as well:
using (var db = new Context())
{
ExistingPais pais = new Pais{Id =theId};
db.Pais.Attach(pais);
db.UF.Attach(editedUF);
editedUF.Pais = pais;
db.Entry(editedUF).State = EntityState.Modified;
db.SaveChanges();
}
Assuming your code looks like this:
public class Pais
{
public int Id { get; set; }
public virtual ICollection<UF> UFs { get; set; }
}
public class UF
{
public int Id { get; set; }
public virtual Pais Pais { get; set; }
public int PaisId { get; set; }
}

Categories

Resources