Model (simplified):
public interface IHaveNotes
{
IList<Note> Notes { get; }
}
public class AnyEntity : IHaveNotes
{
public virtual IList<Note> Notes { get; set; }
}
public class Note
{
public virtual string Text { get; set; }
}
Mapping with components:
<class name="AnEntity">
<id .../>
<list name="Notes" table="AnEntityNote">
<key />
<list-index />
<composite-element class="Note">
<property name="Text" />
</composite-element>
</list>
</class>
That's really simple. Just a table per Notes collection (i.e. one table per
entity supporting Notes).
Mapping with many-to-many:
<class name="AnEntity">
<id .../>
<list name="Notes" table="AnEntityNote" cascade="all,delete-orphan">
<key />
<list-index />
<many-to-many class="Note">
<column name="Note" unique="true"/>
</many-to-many>
</list>
</class>
<class name="Note">
<id .../>
<property name="Text" />
</class>
You need to use many-to-many because that means the relationship is NOT
stored in the Note table, but in the specified relationship table. It
doesn't matter if the relationship is bidirectional or not. The addition of
the unique clause makes sure a note can not be associated with multiple
entities.
In this case, you have a single Note table whichs stores Note data, and a
table per collection which stores keys/indexes only.
I believe this model and implementations will give you the best flexibility
and performance for your usage. Remember that you'll probably want to
see/update "The notes that belong to this entity" and not "The unknown
entity this note is associated to", which is the case for <any>
Diego
On Tue, Jul 13, 2010 at 22:26, Vekaz Turković <[email protected]> wrote:
> Diego, I'm sorry. I must have turned my brain off after reading "you could
> have an interface..." in you first reply. I thought you were suggesting to
> have IHaveNotes mapped and all implementations mapped as subclasses, and
> single table Notes would then reference IHaveNotes. That was the only
> mapping I could come up with and I assumed you were suggesting the same.
>
> This is your original suggestion:
> > - Note is a component type, there is one table per Notes collection
> > - Note is an entity, there is one table for the Notes and a many-to-many
> > table per each collection, mapped with cascade="all-delete-orphan"
> and I don't understand it very well since my knowledge in this area is very
> limited.
> 1. I haven't used Component before, would you care to explain exactly how
> to implement this?
> 2. I don't understand the need for many-to-many; a single Note cannot be
> attached to multiple entities. With that in mind, this would become
> "one-to-many per each collection" and Notes table would simply reference
> IHaveNotes implementations' tables, right? Or am I missing something? (I was
> going to mention performance issues again, but remembered your point #1 and
> got scared :))
>
> I only glanced at description of <any>, but if I my assumptions about it
> are correct, it would allow me to attach a note to anything, I could still
> have IList<Note> where I needed it and navigate Notes via IHaveNotes and
> changes to the model would be slightly easier to implement (i.e. I wouldn't
> have to change the model, the db and the mappings to add or remove an
> IHaveNotes implementation. Just implement the interface and add the type to
> <any>). Am I wrong about that?
>
> I am really thankful for the time you spent helping me with this, and ask
> you to spend just a little bit more :)
>
>
> On Tue, Jul 13, 2010 at 3:18 PM, Diego Mijelshon
> <[email protected]>wrote:
>
>> Vekaz,
>>
>> 1. You first have to try a design before deciding for something more
>> complicated because of performance issues you can't measure or describe
>> 2. There is no inheritance in the proposed model
>> 3. <any> has its own problems, and, in this case in particular, I don't
>> think it's the best fit, as the relationship would be on the Note side, not
>> on the entity, so it's not easy to retrieve all the notes for an entity,
>> whereas the other model allows for navigation.
>> 4. You write the mappings once (and you can get a tool like ConfORM to do
>> most of the heavy lifting), but you use the model in many places. So it's
>> the model what needs to be simple and easy to use, not the mappings (which
>> are there to abstract the complexity of the relational implementation)
>>
>> Diego
>>
>>
>> On Tue, Jul 13, 2010 at 08:44, Vekaz Turković <[email protected]> wrote:
>>
>>> @Peter
>>> Maybe this will help:
>>> http://ayende.com/Blog/archive/2006/06/05/NHibernatesAssociationIsCool.aspx
>>> It's a bit dated but I guess it's still relevant.
>>>
>>> @Diego
>>> Nothing in particular, Diego. I don't know much about db/nhib
>>> performance, but from what I hear, having as many relationships as that
>>> design suggests, on the most frequently used tables, might cause performance
>>> issues. And, IIRC, I read somewhere that when working with inheritance
>>> queries should be given more thought than 'regular' queries, and that design
>>> would have inheritance everywhere. Why chose design that implies more
>>> thinking? :) <any> seems much more simple. Again, that's what I remember
>>> hearing/reading and isn't necessarily true, so... Feel free to educate me ;)
>>>
>>> On Tue, Jul 13, 2010 at 8:37 AM, Kenneth <[email protected]> wrote:
>>>
>>>> Yes, as Fabio wrote, use the <any> tag
>>>>
>>>> /K
>>>>
>>>> On Jul 13, 7:37 am, Peter Forstmeier <[email protected]>
>>>> wrote:
>>>> > Hi all,
>>>> > i follow this thread because i'm very interested on the same subject.
>>>> > I didn't see any solution right now. Did i miss something.
>>>> >
>>>> > Peter
>>>> >
>>>> > On 12 Jul., 13:53, Diego Mijelshon <[email protected]> wrote:
>>>> >
>>>> > > Uh?
>>>> >
>>>> > > What performance problem are you worried about exactly?
>>>> >
>>>> > > Diego
>>>> >
>>>> > > On Sun, Jul 11, 2010 at 22:50, Vekaz Turković <[email protected]>
>>>> wrote:
>>>> > > > Thank you Diego, I was afraid this will end up unanswered :) That
>>>> was one
>>>> > > > way I considered doing it, but there's also Activities who are,
>>>> > > > relationship-wise, the same as notes. Both Notes and Activities
>>>> will most
>>>> > > > probably hold a lot of data. I am afraid that, with so many
>>>> relationships,
>>>> > > > performance might become an issue pretty soon. Would it be
>>>> possible to use a
>>>> > > > denormalized database; Note would hold type and id of the entity
>>>> it is
>>>> > > > attached to?
>>>> >
>>>> > > > On Sun, Jul 11, 2010 at 11:43 PM, Diego Mijelshon <
>>>> [email protected]>wrote:
>>>> >
>>>> > > >> This was discussed last month (
>>>> > > >>
>>>> http://groups.google.com/group/nhusers/browse_thread/thread/973df7dce..
>>>> .)
>>>> > > >> and it is certainly on topic :-)
>>>> >
>>>> > > >> In short: the model is simple. Several classes have a collection
>>>> of notes
>>>> > > >> (or, more precisely, a List, because the order probably matters).
>>>> You could
>>>> > > >> create an interface (IHaveNotes) to deal with this in your code
>>>> and make it
>>>> > > >> easy to add/retrieve notes from any entity.
>>>> >
>>>> > > >> Then, you have at least two possible ways to map this:
>>>> > > >> - Note is a component type, there is one table per Notes
>>>> collection
>>>> > > >> - Note is an entity, there is one table for the Notes and a
>>>> many-to-many
>>>> > > >> table per each collection, mapped with
>>>> cascade="all-delete-orphan"
>>>> >
>>>> > > >> Note that the data implementation should NOT change your model.
>>>> It's
>>>> > > >> always an IList<Note>.
>>>> >
>>>> > > >> Diego
>>>> >
>>>> > > >> On Sun, Jul 11, 2010 at 08:01, moohzor <[email protected]> wrote:
>>>> >
>>>> > > >>> Hi,
>>>> >
>>>> > > >>> This isn't strictly nhibernate related, but I thought this would
>>>> still
>>>> > > >>> be the right place to ask.
>>>> >
>>>> > > >>> I have a Note entity, related to almost all other entities (you
>>>> can
>>>> > > >>> attach a note to pretty much anything else). All these entities
>>>> have a
>>>> > > >>> property to hold a collection of notes (but notes don't have to
>>>> have
>>>> > > >>> reference to the entity they're related to). I don't know how to
>>>> best
>>>> > > >>> model this in the database and map in hbm. How would you do
>>>> this?
>>>> >
>>>> > > >>> Thanks in advance.
>>>> >
>>>> > > >>> --
>>>> > > >>> You received this message because you are subscribed to the
>>>> Google Groups
>>>> > > >>> "nhusers" group.
>>>> > > >>> To post to this group, send email to [email protected].
>>>> > > >>> To unsubscribe from this group, send email to
>>>> > > >>> [email protected]<nhusers%[email protected]>
>>>> <nhusers%[email protected]<nhusers%[email protected]>
>>>> >
>>>> > > >>> .
>>>> > > >>> For more options, visit this group at
>>>> > > >>>http://groups.google.com/group/nhusers?hl=en.
>>>> >
>>>> > > >> --
>>>> > > >> You received this message because you are subscribed to the
>>>> Google Groups
>>>> > > >> "nhusers" group.
>>>> > > >> To post to this group, send email to [email protected].
>>>> > > >> To unsubscribe from this group, send email to
>>>> > > >> [email protected]<nhusers%[email protected]>
>>>> <nhusers%[email protected]<nhusers%[email protected]>
>>>> >
>>>> > > >> .
>>>> > > >> For more options, visit this group at
>>>> > > >>http://groups.google.com/group/nhusers?hl=en.
>>>> >
>>>> > > > --
>>>> > > > You received this message because you are subscribed to the Google
>>>> Groups
>>>> > > > "nhusers" group.
>>>> > > > To post to this group, send email to [email protected].
>>>> > > > To unsubscribe from this group, send email to
>>>> > > > [email protected]<nhusers%[email protected]>
>>>> <nhusers%[email protected]<nhusers%[email protected]>
>>>> >
>>>> > > > .
>>>> > > > For more options, visit this group at
>>>> > > >http://groups.google.com/group/nhusers?hl=en.-Zitierten Text
>>>> ausblenden -
>>>> >
>>>> > > - Zitierten Text anzeigen -
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "nhusers" group.
>>>> To post to this group, send email to [email protected].
>>>> To unsubscribe from this group, send email to
>>>> [email protected]<nhusers%[email protected]>
>>>> .
>>>> For more options, visit this group at
>>>> http://groups.google.com/group/nhusers?hl=en.
>>>>
>>>>
>>> --
>>> You received this message because you are subscribed to the Google Groups
>>> "nhusers" group.
>>> To post to this group, send email to [email protected].
>>> To unsubscribe from this group, send email to
>>> [email protected]<nhusers%[email protected]>
>>> .
>>> For more options, visit this group at
>>> http://groups.google.com/group/nhusers?hl=en.
>>>
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "nhusers" group.
>> To post to this group, send email to [email protected].
>> To unsubscribe from this group, send email to
>> [email protected]<nhusers%[email protected]>
>> .
>> For more options, visit this group at
>> http://groups.google.com/group/nhusers?hl=en.
>>
>
> --
> You received this message because you are subscribed to the Google Groups
> "nhusers" group.
> To post to this group, send email to [email protected].
> To unsubscribe from this group, send email to
> [email protected]<nhusers%[email protected]>
> .
> For more options, visit this group at
> http://groups.google.com/group/nhusers?hl=en.
>
--
You received this message because you are subscribed to the Google Groups
"nhusers" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/nhusers?hl=en.
