I have two trivial tables in postgres 14 schema:
Table 1 rubric_questions:
rubric_version_id : int
rubric_id : int
FK: (rubric_id, rubric_version_id) -> rubric(rubric_id, version_id) /// note it is id THEN version
Table 2 rubrics:
rubric_id : int
rubric_version_id : int
PK: (version_id, rubric_id) /// note it is version THEN id
Obviously it is wrong that order of columns in first table's FK is reverse of second tabel's PK.
However when I use Generate DDL function in DBeaver it returns the following, note the FK definition:
CREATE TABLE eval.rubric_questions (
rubric_version_id int4 NOT NULL,
rubric_id int4 NOT NULL
);
ALTER TABLE eval.rubric_questions ADD CONSTRAINT rubric_questions_fk FOREIGN KEY (rubric_id,rubric_version_id) REFERENCES <?>();
I'm trying to scaffold entity framework code-first, and it chokes on this relationship with:
Could not scaffold the foreign key 'eval.rubric_questions(rubric_version_id,rubric_id)'. A key for 'version_id,rubric_id' was not found in the principal entity type 'Rubric'.
Reversing FK columns fixes the issue.
I understand that I will need to fix the column order, but this issue exists in several places, so I would like to get the scaffold going and then fix these problems as I discover them (I have inherited the schema that I need to use in my new EF-based application).
Question: Is this <?>() syntax documented somewhere, or is it postgresql bug? Is there a work-around without changing the schema?u
Related
Table: A
columns names
--------------------------
PK varchar |datasource
PK int |programid
int |workspaceid
composite pk key of: datasource and programid
Table: B
columns names
---------------------------------------
PK varchar | datasource
PK int | quantitycontractid
int | workspaceid
composite pk key of: datasource and quantitycontractid
I need to make relationship between those tables but using workspaceid and datasource. So i try as usual:
ALTER TABLE A
ADD CONSTRAINT fk_relation
FOREIGN KEY (workspaceid, datasource)
REFERENCES B(workspaceid, datasource)
I am getting following error:
there is no unique constraint matching given keys for referenced table
"B"
You need to add UNIQUE key to B(workspaceid, datasource) before you consider that as a foreign key in table A. This is to ensure a correct one to one or one to many relationships between the two tables.
ALTER TABLE B
ADD CONSTRAINT unq_contraint
UNIQUE KEY (workspaceid, datasource)
The error makes perfect sense to me, your table B doesn't have any unique index or primary key linked to workspaceid. Having said that, your table structure for B looks kinda weird to me. Most databases have a primary key that is an autoincrement and one or more foreign keys linking to other tables. You seem to have made your primary key a combination of multiple foreign keys. While this works, you're gonna have issues like you described and have complicated joins when your query your tables.
Vishal R already answered on how to fix your problem.
What T-SQL DDL is required to create a constraint that ensures that the values in a column in one table are the same as the values in a column in a different table?
I want to do this without using a PK-FK relationship.
The T-SQL DDL at the end of this post is an example of the generic problem that I'm trying to solve.
In this example, I want to know how to add an equality constraint between the two tables that ensures that the set of values in the column:
"PersonMayDriveCar.personName"
is always equal to the set of values in the column
"DriverLicense.personName"
CREATE SCHEMA "Equality Constraint"
GO
CREATE TABLE "Equality Constraint".PersonMayDriveCar
(
carVin nchar(4000) NOT NULL,
personName nchar(70) NOT NULL,
CONSTRAINT PersonMayDriveCar_PK PRIMARY KEY(personName, carVin)
)
GO
CREATE TABLE "Equality Constraint".DriverLicense
(
driverLicenseNr int NOT NULL,
personName nchar(70) NOT NULL,
CONSTRAINT DriverLicense_PK PRIMARY KEY(driverLicenseNr),
CONSTRAINT DriverLicense_UC UNIQUE(personName)
)
GO
I see that you want to maintain referential integrity between the two tables without using a foreign key.
Based on my past experience, I solved such an issue using a trigger.
So you can create a trigger on the DriverLicense table which ensures that any insert or update into the DriverLicense table will be rolled back if the inserted driverLicenseNr doesn't exist in the PersonMayDriveCar table.
You can go through this for a full example:
https://www.mssqltips.com/sqlservertip/4242/sql-server-referential-integrity-without-foreign-keys/
Adhere to convention:
Use an FK. It’s that simple.
Don’t link these table together with an FK, because they are both child tables of ...
Create a person table, which is the parent of the other two tables
Try this:
Person
- id (PK)
- name
- other columns
PersonMayDriveCar
- person_id (FK to person)
- other columns
DriverLicense
- person_id (FK to person)
- other columns
We have an existing database which we'd like to connect to using code-first EF.
An issue we have come across is in the following scenario...
**TABLE 1**
DebtorID Int PK (identity column)
Type Int PK FK ( value would be '1' - FK to Table 4)
**TABLE 2**
CreditorID Int PK (identity column)
Type Int PK FK (value would be '2' - FK to Table 4)
**TABLE 3**
TransactionID Int PK
ParentID Int FK
ParentType Int FK
-----------------------
**TABLE 4**
Type Int PK (identity column)
Description Varchar(max)
Table 3 has foreign key links to multiple tables (Table 1, Table 2, etc...). The key is unique by the type column. Constraints have been altered here to allow this scenario.
Q. How can we implement this schema using code-first? We have looked at Data Annotations, and also Fluent API, but can't come up with a solution other than to alter the database schema (which would be a big headache).
As it stands, we only seem to be able to link to a single foreign table.
The solution also has to account for other tables (not shown here) where the primary keys for Table 1 etc... are part of the primary key.
A real world view of our schema, and the issue with code-first.
Current schema:
http://oi60.tinypic.com/2aeug4m.jpg
Code-First schema:
http://oi62.tinypic.com/9schmc.jpg
I have found the answer using the Fluent Api - was much easier than I thought.
Simply in the OnModelCreating() method, I needed to add multiple ForeignKey entries to cover each link.
modelBuilder.Entity(Of TABLE_3).HasRequired(Function(p) p.TABLE_1).WithMany().HasForeignKey(Function(p) New With {p.ParentID, p.ParentType})
modelBuilder.Entity(Of TABLE_3).HasRequired(Function(p) p.TABLE_2).WithMany().HasForeignKey(Function(p) New With {p.ParentID, p.ParentType})
This was in addition to any configuration I did using Data Annotation with regards to setting up primary keys and navigational properties (e.g. navigation properties for TABLE_1 and TABLE_2)
I have a WCF Data Service (5.5) sitting over an EF (5.0) model,
I'm getting the following error when I query $metadata:
"An IEdmModel instance was found that failed validation. The following errors were reported:
InvalidMultiplicityOfDependentEnd : The multiplicity of the dependent end 'QuestionsetMember' is not valid. Because the dependent properties don't represent the dependent end key, the the multiplicity of the dependent end must be '*'."
QuestionsetMember has a composite primary key of 2 columns, each of which is hooked to a primary key of another table, i.e. a foreign key exist from each column of the key to the two tables' primary keys.
I've searched but cannot find any info on "InvalidMultiplicityOfDependentEnd".
Also tried fiddling with the relationships in the EDMX, but changing the End Multiplicity causes errors which won't allow the model to compile.
Any ideas how to get round this (hopefully without changing my schema) ?
This seems to be a very rare error. I did not find anywhere else an explanation of that error. So i did find for me a solution after inspecting every single constraint and every column in both tables. To my great surprise the order of the primary key columns seems to be relevant.
For explanation: I do the database-design within the sql server management studion, and update my model with the entity framework designer.
First Table:
CREATE TABLE Table1
(
Column1 int NOT NULL,
Column2 int NOT NULL,
PRIMARY KEY (Column1,Column2)
)
Second Table:
CREATE TABLE Table2
(
Column1 int NOT NULL,
Column2 int NOT NULL,
PRIMARY KEY (Column1,Column2)
FOREIGN KEY (Column1,Column2) REFERENCES Table1(Column1,Column2)
)
This would work. But it do not work, if you would define the columns of the primary key in the second table in another order:
CREATE TABLE Table2
(
-- Changed order in definition:
Column1 int NOT NULL,
Column2 int NOT NULL,
-- Changed order in PK group:
PRIMARY KEY (Column2,Column1)
-- Leave the FK definition untouched:
FOREIGN KEY (Column1,Column2) REFERENCES Table1(Column1,Column2)
)
I think the order of the column definition has impact on the generated model. And this order could maybe have an impact in the model validation within the IEdmModel class. Who knows...
In EF Code First, we can create one-to-one relationship by coding like this:
public class User
{
public int UserID {get;set;}
public string Name {get;set;}
public int UserDetailID {get;set;}
public UserDetail Detail {get;set;}
}
public class UserDetail
{
public int UserDetailID {get;set;}
public string Address {get;set:}
public int UserID {get;set;}
public User User {get;set;}
}
However, when I tried to create the same relationship by using EF Database first in visual studio 2012, I got in trouble. Here is my code:
CREATE TABLE [dbo].[Users] (
[UserID] UNIQUEIDENTIFIER CONSTRAINT [DF_Users_UserID] DEFAULT (newid()) NOT NULL,
[UserDetailID] UNIQUEIDENTIFIER NOT NULL,
[Name] NVARCHAR (50) NOT NULL,
CONSTRAINT [PK_Users] PRIMARY KEY CLUSTERED ([UserID] ASC),
CONSTRAINT [FK_Users_UserDetails] FOREIGN KEY ([UserDetailID]) REFERENCES [UserDetails]([UserDetailID])
);
CREATE TABLE [dbo].UserDetails] (
[UserDetailID] UNIQUEIDENTIFIER CONSTRAINT [DF_UserDetails_UserDetailID] DEFAULT (newid()) NOT NULL,
[UserID] UNIQUEIDENTIFIER NOT NULL,
[Address] NVARCHAR(100) NOT NULL,
CONSTRAINT [PK_UserDetails] PRIMARY KEY CLUSTERED ([UserDetailID] ASC),
CONSTRAINT [FK_UserDetails_Users] FOREIGN KEY ([UserID]) REFERENCES [dbo].[Users] ([UserID])
The error message is something like
"Error 2 SQL01767: Foreign key 'FK_Users_UserDetails' references invalid table 'UserDetails'.
I think the reason for this error probably be when it tries to reference the foreign key "UserDetailID", it finds that it hasn't been created yet. But I don't know how to fix this, and I don't even know this is the way to do it, I know doing one-to-one relationship with EF is tricky, or some people even says it's impossible. Can anyone give me any suggestion? Thank you.
Update: Just to clarify my case, I am trying to design the database in visual studio 2012 database project, then publish it to the SQL server, afterward, create/update my .edmx file from the database in SQL server. I am not sure about how to create a one-to-one relationship that the EF can recognize correctly and create the right classes in .edmx file.
Creating a 1:1 relationship is not that tricky and certainly not impossible, although it is not a particularly common requirement and in this case I can't see why you would want it? If people are saying this then you are talking to the wrong people.
Anyhow using SQL queries as you seem to be is not to do with EF, you are just working directly with the database, In the first CREATE you are trying to add the constraint but you haven't created the other table yet... As you mentioned in your question.
I think you need to create both tables first and then add the constraint with ALTER TABLE.
Additionally searching SO for questions about 1:1 turns up quite a lot so I suggest you do that.
EDIT: So using a database project (I only have VS Express so I don't have those) you want to create a "1:1" relationship using SQL and then add an Entity Data Model to a (probably different) project which references the database and automatically create 1:1 relationship?
That is a whole different story unfortunately. When I was talking about possibility to create 1:1 that was in reference to EF only and not to databases as such. It is actually very difficult/impossible as you said to create 1:1 in SQL. I think that it makes sense that in order to insert into a 1:1 realationship you would need to somehow insert into both tables at exactly the same time or fiddle about with disabling constraints briefly when adding rows.
In general there are a few different option.
Don't split the tables unnecessarily. In true 1:1 all data is required so the only reason to split is for performance reasons (e.g partioning) which I would avoid in this case.
Map multiple table to a single entity as show here.
Create a 1:0..1 relationship and enforce you own requirements in the application.
In either option 2 or 3 you can use the following SQL to create a relationship which uses the same PK on the second table as the FK in the relationship.
CREATE TABLE [dbo].[Users] (
[UserID] UNIQUEIDENTIFIER CONSTRAINT [DF_Users_UserID] DEFAULT (newid()) NOT NULL,
[Name] NVARCHAR (50) NOT NULL,
CONSTRAINT [PK_Users] PRIMARY KEY CLUSTERED ([UserID] ASC),
);
CREATE TABLE [dbo].[UserDetails] (
[UserID] UNIQUEIDENTIFIER NOT NULL,
[Address] NVARCHAR(100) NOT NULL,
CONSTRAINT [PK_UserDetails] PRIMARY KEY CLUSTERED ([UserID] ASC),
CONSTRAINT [FK_UserDetails_Users] FOREIGN KEY ([UserID]) REFERENCES [dbo].[Users] ([UserID]) ON DELETE CASCADE
);
I suggest you also use store generated identity as well where you can.
Just remove UserDetailID from the UserDetail table and make UserID both primary key and a foreign key to the UserID column of the User table.
This the correct way to make 1:1 relationships in a database and EF recognizes it and maps the entities appropriately with database-first approach.
The question is a couple years old.. and the ef version wasn't stated.. but one answer is to remove UserDetailID from both tables. UserID should be the only primary key on both tables.
the 'unqieidentifier' (GUID) data type shouldn't pose an issue (opposed to using INT), but you certainly don't want to populate it with newId..