I'm very new to NHibernate, and I'm running into a problem while saving a list of child objects.
NOTE
<class name="Note" table="NOTE">
<id name="NoteID" column="NOTE_ID">
<generator class="identity" />
</id>
...
<list name="Sections" table="NOTE_SECTIONS" cascade="all" lazy="false">
<key column="NOTE_ID"/>
<index column="SORT_ORDER"/>
<one-to-many class="Section"/>
</list>
</class>
NOTE SECTION
<class name="Section" table="NOTE_SECTIONS">
<id name="SectionID" column="Section_ID">
<generator class="identity" />
</id>
<property name="NoteID" column="NOTE_ID"/>
...
</class>
The mappings work perfectly for reading the data. However, when I make a change to the Note Section, The queries it generates appears to be going through the proper steps, but then I get the following error:
NHibernate.Exceptions.GenericADOException: could not delete collection: [Domain.Note.Sections#1][SQL: UPDATE NOTE_SECTIONS SET NOTE_ID = null, SORT_ORDER = null WHERE NOTE_ID = #p0] ---> System.Data.SqlClient.SqlException: Cannot insert the value NULL into column 'NOTE_ID', table 'NOTE_SECTIONS'; column does not allow nulls. UPDATE fails.
I have read that in order to save like this it will need to be bidirectional. But I've also read that bidrectional mappings don't work with Lists. It's important that my collection have a maintained order - what's the best way to save?
You should use inverse="true" on your collection mapping if you wish to save child objects in this manner.
<list name="Sections" table="NOTE_SECTIONS" inverse="true" cascade="all" lazy="false">
<key column="NOTE_ID"/>
<index column="SORT_ORDER"/>
<one-to-many class="Section"/>
</list>
Inverse Attribute in NHibernate
probably you would need to refer Note from Note Section as many-to-one relation.
Related
I'm trying to map the following classes:
PessoaFisica and PessoaJuridica inherits Pessoa.
Cliente has an association with Pessoa, it may be PessoaJuridica or PessoaFisica.
When I save a Cliente object with PessoaFisica, for example, thats ok. But when I try to update and I set the property Pessoa from Cliente to PessoaJuridica and try to update, it updates, but it generates a new row in table TB_PESSOA and the old row, in PessoaFisica is not deleted. It creates a new row to PessoaJuridica, but the old row remains. What's wrong with my mapping XMLs ? Why NHibernate does not delete the old row before insert the new polymorphic object ?
Those are the mapping files I am using
<hibernate-mapping xmlns="urn:nhibernate-mapping-2.2"
namespace="SALClassLib.Masterdata.Model" assembly="SALClassLib">
<class name="Pessoa" table="TB_PESSOA">
<id name="Id">
<column name="ID_PESSOA" not-null="true"/>
<generator class="increment" />
</id>
(other properties...)
<joined-subclass name="PessoaFisica" table="TB_PESSOA_FISICA">
<key column="ID_PESSOA" />
(other properties...)
</joined-subclass>
<joined-subclass name="PessoaJuridica" table="TB_PESSOA_JURIDICA">
<key column="ID_PESSOA" />
(other properties...)
</joined-subclass>
</class>
<class name="Cliente" table="TB_CLIENTE">
<id name="Id">
<column name="ID_CLIENTE" not-null="true"/>
<generator class="increment" />
</id>
<many-to-one name="Pessoa" class="Pessoa" cascade="all" column="ID_PESSOA" not-null="true" unique="true" />
Thank you
NHibernate cascading is nicely explained here: NHibernate Cascades: the different between all, all-delete-orphans and save-update
One of the option, is cascade="all-delete-orphan" which could be seen as what you are asking for.
BUT
Cascading deletion of the orphans is correct only in parent-child scenario (no parent ==> no children) or one-to-one mapping. (i.e not vice versa child-parent)
In your case, you do ask for deletion of the referenced object. But NHibernate (well no-one) can know, if it is not referenced by some other "child".
If you need to delete previous Person assigned, you can always do it in code - but explicitly
I've got a scheme (fields aren't necessary):
a busy cat http://picsearch.ru/share/image-BCE8_4E168F3B.jpg
I've got mappings:
Entity
<class name="LogicalModel.Entity" table="`Entity`" lazy="true">
<id name="Id" ..> ... </id>
<bag name="Attributes" lazy="true" cascade="all-delete-orphan" fetch="select" batch-size="1" access="property" inverse="true">
<key column="`Entity`" />
<one-to-many class="LogicalModel.Attribute" />
</bag>
<bag name="Keys" lazy="true" cascade="all-delete-orphan" fetch="select" batch-size="1" access="property" inverse="true">
<key column="`Entity`" />
<one-to-many class="LogicalModel.Key" />
</bag>
</class>
Attribute
<class name="LogicalModel.Attribute" table="`Attribute`" lazy="true">
<id name="Id" ..> ... </id>
<many-to-one name="Type" class="LogicalModel.Entity" column="`Type`" cascade="save-update" fetch="select" not-null="true" foreign-key="fk_TypeAttribute" />
<many-to-one name="Entity" class="LogicalModel.Entity" column="`Entity`" cascade="none" fetch="select" not-null="true" foreign-key="fk_EntityAttributes" />
</class>
Key
<class name="LogicalModel.Key" table="`Key`" lazy="true">
<id name="Id" ..> ... </id>
<bag name="KeyAttributes" lazy="true" cascade="all-delete-orphan" fetch="select" access="property" inverse="true">
<key column="`Key`" />
<one-to-many class="LogicalModel.KeyAttribute" />
</bag>
<many-to-one name="Entity" class="LogicalModel.Entity" column="`Entity`" cascade="none" fetch="select" not-null="true" foreign-key="fk_EntityKeys" />
</class>
KeyAttribute:
<class name="LogicalModel.KeyAttribute" table="`KeyAttribute`" lazy="false">
<id name="Id" ..> ... </id>
<many-to-one name="Attribute" class="LogicalModel.Attribute" column="`Attribute`" cascade="save-update" fetch="select" not-null="true" foreign-key="fk_AttributeKeyAttribute" />
<many-to-one name="Key" class="LogicalModel.Key" column="`Key`" cascade="none" fetch="select" not-null="true" foreign-key="fk_KeyKeyAttributes" />
</class>
Now please take a look...
As you see, We've got one-way master association KeyAttribute - Attribute, so it's just many-to-one and I don't need back association at all.
Now the problem is when I'm trying to delete whole graph - delete Entity object (notice: Entity actually aren't loaded at all, it's just set of proxies, that's why NHibernate make additional SELECT queries to check references before delete)
like this
Session.Delete(Entity); // here PropertyValueException:
// not-null property references a null or transient value: LogicalModel.KeyAttribute.Attribute
Session.Flush(); // Actually I use transactions in my code, but don't mind
SQL Profiler:
exec sp_executesql N'SELECT entities0_.[Id] as Id1_1_, entities0_.[Id] as Id1_45_0_,
FROM [Entity] entities0_ WHERE entities0_.[LogicalModel]=#p0',N'#p0 uniqueidentifier',#p0='DC8F8460-9C41-438A-8334-97D0A94E2528'
exec sp_executesql N'SELECT attributes0_.[Entity] as Entity12_1_, attributes0_.[Id] as Id1_1_, attributes0_.[Id] as Id1_16_0_, attributes0_.[Type] as Type11_16_0_, attributes0_.[Entity] as Entity12_16_0_
FROM [Attribute] attributes0_ WHERE attributes0_.[Entity]=#p0',N'#p0 uniqueidentifier',#p0='63E4D568-EAB2-4DF2-8FED-014C8CB2DE22'
exec sp_executesql N'SELECT keys0_.[Entity] as Entity4_1_, keys0_.[Id] as Id1_1_, keys0_.[Id] as Id1_43_0_, keys0_.[Entity] as Entity4_43_0_
FROM [Key] keys0_ WHERE keys0_.[Entity]=#p0',N'#p0 uniqueidentifier',#p0='63E4D568-EAB2-4DF2-8FED-014C8CB2DE22'
exec sp_executesql N'SELECT keyattribu0_.[Key] as Key4_1_, keyattribu0_.[Id] as Id1_1_, keyattribu0_.[Id] as Id1_0_0_, keyattribu0_.[Attribute] as Attribute3_0_0_, keyattribu0_.[Key] as Key4_0_0_
FROM [KeyAttribute] keyattribu0_ WHERE keyattribu0_.[Key]=#p0',N'#p0 uniqueidentifier',#p0='103D8FB3-0B17-4F51-8AEF-9623616AE282'
So what we can see:
not-null property references a null or transient value: LogicalModel.KeyAttribute.Attribute
happened just after NH check field Attribute (not-null constraint in db, it's ok) in class KeyAttribute (see profiler log).
It's pretty fun, cause NH have to delete Attributes and KeyAttributes both, NH read information about Attribute field in KeyAttribute class, FOUND it in DB, NOT FOUND it in NH session (!!!) (cause Attributes was loaded before), and just throw this stupid error.
What I've already tried to do:
1. make not-null="false". In this case NH makes additional update - try to set Attribute=NULL - cause constraint violation in DB.
2. set lazy="false", lazy="no-proxy" on many-to-one association for KeyAttribute-Attribute - nothing;
Now I don't like the idea of interceptors because there are to many scenarios where I've got the same situation, I need common solution
Please, guys, any suggestions?
In my opinion it may be caused by your lazy load on all entities of model.
When deleting entity, it loads and delete referenced Attribute list, loads referenced Key list, loads referenced KeyAttribute list (to have key of deletion) and then it falls in not-null property references a null or transient value because referenced Attribute has been deleted before in session.
You can check that by removing all lazy load in your mapping files.
A quick solution may be to keep lazy load but to force a full load of model (with hibernate initialize()) when deleting, for example in a Delete(Entity) static method in Entity factory.
Have you tried setting on-delete="cascade" in
<class name="LogicalModel.Key" table="`Key`" lazy="true">
<id name="Id" ..> ... </id>
<bag name="KeyAttributes" lazy="true" cascade="all-delete-orphan" fetch="select" access="property" inverse="true">
<key column="`Key`" on-delete="cascade" />
<one-to-many class="LogicalModel.KeyAttribute" />
</bag>
<many-to-one name="Entity" class="LogicalModel.Entity" column="`Entity`" cascade="none" fetch="select" not-null="true" foreign-key="fk_EntityKeys" />
Because in profile you will see nh trying to update something to null which is non nullable
NH sometimes requires to set references to null. Usually this is to avoid problems in models where circular references exist. But it is not always clever enough to find a way to avoid it, even if the is one.
So it may require to allow nulls in some foreign key fields, of course not only in the mapping file, also in the database. It actually should solve the problem.
Alternatively, you could also delete the data table by table using HQL. This works fine in all cases where you don't have inheritance and if you know all entities and the order to delete them:
object entityId;
// gets keys to delete
List<object> keyIds = Session
.CreateQuery("select id from Key where Entity = :entity")
.SetEntity("entity", Entity)
.List<object>();
// delete KeyAttribute which reference the key
Session.CreateQuery("delete KeyAttribute where Key.id in (:keyIds)")
.SetParameterList("keyIds", keyIds)
.ExecuteUpdate();
// delete the keys
Session.CreateQuery("delete Key where id in (:keyIds)")
.SetParameterList("keyIds", keyIds)
.ExecuteUpdate();
// get attributes to delete
List<object> attributeIds = Session
.CreateQuery("select id from Attribute where Entity = :entity")
.SetEntity("entity", Entity)
.List<object>();
// delete KeyAttributes which reference the attributes
Session.CreateQuery("delete KeyAttribute where Attribute.id in (:attributeIds)")
.SetParameterList("attributeIds", attributeIds )
.ExecuteUpdate();
// delete the attributes
Session.CreateQuery("delete Attribute where id in (:attributeIds)")
.SetParameterList("attributeIds", attributeIds )
.ExecuteUpdate();
Session.CreateQuery("delete Entity where id = :entityId")
.SetParameter("entityId", Entity.Id)
.ExecuteUpdate();
Note:
You may break the parameter lists into piece if they exceed the size of around 2000 (in SQL Server).
The session gets out of synch when deleting directly in the database. This doesn't cause any problems when deleting is all you do. When you are doing other staff in the same session, clear the session after deleting.
Hey all, I'm kicking the tires on NHibernate and have a conoundrum I have been scratching my head over for a bit now, working with a legacy database with some fairly complex relationships.
ClaimRoot has a primary key of a claimGUID.
ClaimRoot has a bag of Claimdetails associated by claimGUID (this works a treat).
The problem is that ClaimRoot also has an optional one to one relationship with ClaimFinancials (not all ClaimRoots have ClaimFinancials, but most do). But the PK for ClaimFinancials is a FormID field. This field exists in the ClaimRoot, but is not the PK.
I've posted a mapping below with extra columns removed to protect the innocent.
<class name="ClaimRoot" table="tbl_ClaimRoot" schema="DB1.dbo">
<id name="ClaimGUID">
<generator class="guid"/>
</id>
<property name="FormID" />
<property name="LastFormNoteText" />
<bag name="ClaimDetails" inverse="true">
<key column="ClaimGUID"/>
<one-to-many class="ClaimDetails"/>
</bag>
</class>
<class name="ClaimDetails" table="tbl_ClaimDetails" schema="DB2.dbo">
<id name="RowID">
<generator class="native"/>
</id>
<property name="ClaimGUID" />
<property name="SeqNo"/>
<property name="B1A_InsID" />
<many-to-one name="Root" column="ClaimGUID" foreign-key="ClaimGUID"/>
</class>
<class name="ClaimFinancials" table="tbl_ClaimFinancials" schema="DB1.dbo">
<id name="FormID">
<generator class="native"/>
</id>
<property name="CreatedDate"/>
<property name="SubmittedDate" />
</class>
Thanks in advance!
-Bob
Assuming the FormID is use only for linking ClaimRoot and ClaimFinancials, it sounds like you want a many-to-one relationship from ClaimRoot to ClaimFinancials. Replace the FormId property on ClaimRoot with a many-to-one.
<class name="ClaimRoot" table="tbl_ClaimRoot" schema="DB1.dbo">
...
<many-to-one name="ClaimFinancials" column="FormID" />
...
</class>
A many-to-one relationship can be be used even if there is only 'one' on the 'many' side. If you were generating a schema, you can specify unique="true" to generate the constraint in the database. With a legacy database, that won't matter.
It is possible to specify an arbitrary SQL where clause for collection mappings. For example:
<map name="myEntity" where="foo = 1" />
However if the column name is ambiguous for some reason, the sql fails. For example, this can occur if you are trying to use joins for example.
Given that the table aliases are automatically generated, you can't qualify the column name. This makes the feature seem rather silly. Does anyone know if there is a work around?
NHibernate should figure out the correct alias for the property you are referencing. Is foo a mapped property of the item entity type (the item type that is in the map collection) ?
For example this works:
<class name="Category" table="Category">
<id name="Id">
<generator class="guid.comb" />
</id>
<property name="Name" not-null="true" length="255" />
<bag name="ProductList" table="Product" cascade="none" where="Name like '%test%'" fetch="join">
<key column="CategoryId" />
<one-to-many class="Product" />
</bag>
</class>
There is a property on both Category and the Product class named "Name" but nhibernate will in this case use the on defined on the Product class.
I have a Group. That group can contain groups and this needs to be mapped with NHibernate. I have done this before but this time I got an existing database that does not really store the data like I would like it too.
The database looks like this:
Groups
Id
Groups_Link
ParentId
ChildId
I have no clue about how to map this?
Edit:
Seems like it's not that easy. One group can exist as a child to more than one group. So it should be an many-to-many association with itself I guess. Well, have yet to figure out how to do this. Any hints would be really appreciated!
Edit:
I thought views could help me solve the problem but there seems to be some restrictions on insert, update and delete with views that makes it difficult.
As you say, it's just a many-to-many that happens to have the same class on both sides of the join.
So:
<class name="G.Group, G" table="GROUPS">
<id column="Id" name="Id">
<generator class="guid"/>
</id>
<set name="Parents" table="Groups_Link" cascade="save-update">
<key column="ChildId"/>
<many-to-many class="G.Group, G" column="ParentId"/>
</set>
<set name="Children" table="Groups_Link" cascade="save-update" inverse="true">
<key column="ParentId" />
<many-to-many class="G.Group, G" column="ChildId"/>
</set>
</class>
I mapped a similar structure like:
<class name="Folder" lazy="false">
<id column="Id" name="Id">
<generator class="guid"/>
</id>
<property name="Name" column="FolderText"></property>
<bag name="Children" cascade="all" fetch="subselect" inverse="true">
<key column="ParentId" />
<one-to-many class="Folder" />
</bag>
<many-to-one class="Folder" name="Parent" column="ParentId" insert="true" />
Edit
I am taking a total shot in the dark now since your schema is not what I've used, and now I understand why you asked this:-) Anyways here's is a thought. Please let us know if this worked of if you find another solution:
<join fetch="join" table="GroupLinks">
<key column="ChildId"></key>
<many-to-one name="Parent" class="Folder" column="ParentId"/>
</join>
I might have the relationship backwards, but this at least would get you the parent.