I am trying to create a strongly typed projection entity "ProjectedPersons" via the designer in the Entity Framework.
I am using POCO classes and have seperated them out into seperate projects.
I keep getting an error:
error 3027: No mapping specified for the following EntitySet/AssociationSet - ProjectedPersons.
How can I remove this error. I only want to use the entity for projections not mapping to the database.
Thanks
If you are using the object only for projections - I would recommend creating it as a complex type instead. You get much of the same features without a lot of the contextual overhead that is inherent to an entity. I believe an entity is fully expecting to be mapped to a database feature in some respect - hence your error.
If you want to define projection in designer you must use QueryView (projection query defined directly in the designer and mapped to the new readonly entity) - that is the feature which must be manually defined directly in XML (MSL part) not in the designer. Otherwise you must do projection manually in Linq query. For linq query you don't need to create entity in the designer. You can use any class which is not mapped in the designer.
The QueryView was the solution. I found a helpful example.
http://msdn.microsoft.com/en-us/magazine/ee336312.aspx
Related
I'm using Self Tracking Entities that implements IObjectWithChangeTracker with the last Entity Framework RC available as a Nuget. The target database is PostgreSQL. I'm also using Code First fluent API to construct the model and LINQ to Entity for querying the database.
To my surpise, a simple SELECT query on the entity generates a SQL query with a mysterious column ChangeTracker_ChangeTrackingEnabled that does not exist in the datatable ! I do not understand this behavior as it seems to me that the EntityTypeConfiguration derived class maps the entity properties to the datatable columns in its constructor.
Is there a way to disable this behavior or at least tell which column should be mapped by the change tracker ?
For that purpose, Context.Configuration.AutoDetectChangesEnabled = false or calling IsConcurrencyToken() mapping in the EntityTypeConfiguration derived object does not help.
Any help appreciated.
TIA.
You must inform EF about every public property you want to avoid in mapping by either marking property with NotMapped attribute or by using Ignore in fluent API.
Btw. as I know STEs are not designed to be used with code first or DbContext API.
I have an entity/table that uses sqlgeography.
Since EF 4.X doesn't support spatial types I'm instead sending the bytes of the field back and forth.
I have stored procs on the database side that handles the converstion and properties on the code side to do that job.
To add the properties in the code I used a partial class.
One of those properties is for the SqlGeography which simply wraps around the byte[] property to handle getting and setting.
This property is hidden from EF using the NotMappedAttribute.
The other is the property exposing the byte[] itself and is decorated with the EdmScalarPropertyAttribute and DataMemberAttribute.
I then go to the EF model designer (*.edmx) to point the entity model at the Insert/Update/Delete stored procs.
It finds the stored procs alright and realises that they (when appropriate) take a VARBINARY parameter.
It also has a drop down allowing you to select a property on the entity class which maps to that parameter.
However this drop down doesn't list either of my properties. I don't care about the SqlGeography property since that is meant to be hidden from EF, however it is vital for me to be able to point it at the byte[] property, as that is where the data comes from.
I would very much like to avoid database triggers or wrapper classes and addiitonal fields to fudge this in to working.
I tried manually editing the .edmx file to include the byte[] property, but then it just complains it's unmapped.
Can anyone give me some insight in to how to get this to work? Or an alternative method of achiving the end result?
We could use a view to create the binary field for us, but this then involves manually creating a lot of the xml for the relationships within the data.
This pretty much voids the point of using EF which is to make life simple and easy.
For this project We'll just add a binary field to the table then have sprocs to handle the converstion on the server and a property in a partial entity class for exposing the geography type in the model.
Next project I doubt we'll be using EF. Dapper is so much more painless, even if theres a touch more code writing involved.
Here's the links for using views if anyone thinks it would be applicable to them:
http://thedatafarm.com/blog/data-access/yes-you-can-read-and-probably-write-spatial-data-with-entity-framework/
http://smehrozalam.wordpress.com/2009/08/12/entity-framework-creating-a-model-using-views-instead-of-tables/
In the end we created a computed column for each table that exposes the spatial data as bytes.
We then use stored procs for inserting and updating the spatial data.
I have an Entity-class having a Property of type Int32: on generating DDL using DevArt for ORACLE a NUMBER(10) column is generated. Reading and writing instances works flawlessly.
However, on fetching instances of this Entity-class sending a custom query to ExecuteStoreQuery on the ObjectContext this Property seems to be returned as System.Double, as such constructing the instances fails.
Can I hint DevArt to construct System.Int32?
Thank you.
Bart
The reason is the fact that OracleDataReader, which is used in the ExecuteStoreQuery method, has type mapping different from the one used in the Entity Framework provider.
I recommend you to use NumberMappings, I suppose you will need to map Number(10) to Int32: Number Mappings=((NUMBER,10,10,System.Int32). These changes should be persisted to the model connection string (they are duplicating the default EF mapping rules, it is necessary for the OracleDataReader from ExecuteStoreQuery).
Please let us know if the problem persists.
I'm building an application using MVC 3 and Entity Framework 4.
I've created my Entity Data Model and generated a database from it.
Now I know the validation attributes such as [Required] or [StringLength(5)] can be used on the model properties to provide validation both clientside and serverside.
I would like to know if these attributes can also be generated dynamically instead of having to add them to the model explicitly? I saw that in EF 4.1 RC you can make use of the Fluent API to further configure your model in the OnModelCreating method by using the DbModelBuilder class.
As shown here
I'm working with a framework however that still uses ObjectContext instead of DbContext so I would like to know if the above solution can be used in combination with ObjectContext?
As a final note, since I've been trying to figure out how to generate and use data annotations it seems using view models would increase the complexity of validation. From what I read here it seems that just passing the models directly to the view would remove the need to add annotations to the models as well as the view models. However that means that you can no longer use strongly typed views when you do joins on the models and pass those to the view directly?
No it can't. Fluent API is different approach to describe mapping. You can use fluent API or EDMX (Entity Data Model). Not both. Fluent API also works only with DbContext API. If you want to have annotations generated you can try to modify T4 template generating your classes.
I have come across a disturbing issue when using poco classes that are extending base classes.
For example, let say you have a Person poco class that has a strongly typed Car property. You also have a Spouse poco that also uses the Car Property.
Now you want to display "Person Car" and "Spouses Car" in the view using the Display("Name = xxx") attribute. You cant!!! Becareful of this issue if you are not using flat View Models
I am using Entity Framework and generating my POCO classes via T4 - these classes inherit from nothing and are very plain and simple (created via template in vs 2010)
I tried using the Metadatatype Attribute so I could create a buddy class but when I did this I no longer was able to see my properties... if I removed the attribute! the properties appeared.
Anyway, searching deeper I found this statement from Microsoft
The associated class must be used with EDM or LINQ-to-SQL models because CLR types cannot mark existing properties with new attributes. If you are working with CLR objects directly, sometimes referred to as Plain Old CLR Object (POCO) types, you can apply the attributes directly to the model
So it appears it doesn't work. Anyway it's difficult for me to insert my Data Annotation on the MODEL itself because it's created via T4 hence if I edit it and then re-run the tool it will remove all my changes.
There is a pretty strong consensus around SO and the MVC blogosphere that you shouldn't annotate your business/crud/domain classes with attributes. Not only is your entire MVC stack becoming dependent upon your business/database classes but you'll quickly end up with multiple context scenarios ( same Model, different validation rules ) that are impossible to validate with just a single model.
Use separate view models for your screens, annotate those.
Based on your comment: "Data Annotation on the MODEL itself because its created via T4 hence"
What I'm trying to say is put your dataannotations on your viewmodels, leave your POCO models alone.