Thank you all for the replies,

in fact most of them scare me a bit because I'm using assigned GUID's
and I see that you discourage this practice... I guess that all
already persisted entities are previously loaded but I clearly
identify from your replies that the tree entities construction in my
application has some flaws.


BTW if I use generated GUID like with comb strategy after every insert
NH will need to query again to retrieve this generated field and at
the end I will need another select for each insert. Right?


Regards,

Guillem Solà

On Mar 20, 2:28 pm, Ramon Smits <[email protected]> wrote:
> You can use Session.Save(..) for the newly created objects. This way you
> force NHibernate to do an INSERT. The problem is that sometimes this fails
> because a root/parent objects does not yet exists in either the database or
> set to the objects properties. In that case you should
>
>  * take the lookup for granted as NHibernate does not know if it needs to
> insert or update.
>  * don't use assigned identifiers
>  * remove the explicit relation between the child and the parent and don't
> use treat it like an aggregate
>
> Regarding the 2nd, don't use assigned identifiers, I learned the hard way
> that assigned identifiers should not be used as primary keys within the
> database to share as it prevents optimized saving of aggregates in
> nhibernate. This conflicts with modern principles that you want to generate
> a key upfront to correlate different actions to. When you need such keys
> then you should keep the primary key internal/protected so that it does not
> have assigned identifiers but use for example hilo and have a guid property
> as a 'natural' key.What I dislike about this is that you will always need
> to join between the parent/child to fetch the child records. If that key
> was part of the primary key then you could just do a simple select.
>
> Problem is that in most situations you would want to say to nhibernate
> 'treat proxies as updates and non proxies as inserts'. That would prevent
> such lookups.
>
> Hope this helps!
>
> On Tue, Mar 20, 2012 at 12:34 PM, Richard Brown 
> <[email protected]>wrote:
>
>
>
>
>
>
>
>
>
> > If you use assigned-identifier, and mix 'old'+'new' (i.e.,
> > persistent+transient) objects, NH cannot determine what has to be inserted
> > and what is already there just by looking at the IDs.
>
> > You should load the existing persisted entities first.
>
> > var referenceEntity =
> > session.Load<ReferenceType>(idThatIKnowExistsInTheDb);
>
> > The best solution is to not use assigned identifiers.
>
> > On 19 March 2012 21:26, guilemsola <[email protected]> wrote:
>
> >> Hi all,
>
> >> I have discovered that with nhibernate profiler with entities created
> >> for the very first time...
>
> >> In my application I create a new entity that has a list of steps, and
> >> each step a list of values. (I'm using inverses on mappings). Also
> >> entities and steps references to master values already in db. So there
> >> is a kind of mix of old and new objects.
>
> >> When I do the first Save I do Session.Save(entity) and the whole tree
> >> is saved in database as it should be, but NH profiler warns about
> >> that:
>
> >> Unable to determine if StepValueEntity with assigned identifier
> >> ede6a5ee-b4bd-4f67-9c64-11ef85b7d6ff is transient or detached;
> >> querying the database. Use explicit Save() or Update() in session to
> >> prevent this.
>
> >> And effectively prior all the new entities insert there are a lot of
> >> useless selects from NH because entities are not on DB.
>
> >> This is very inefficient, so what is the correct way to store those
> >> new entities and tell NH that they are new?
>
> >> FYI this is how I do mapping for identity columns, maybe this doesn't
> >> give a clue to nhibernate to know about what are new and already
> >> persisted entities and I should do it in another way.
>
> >> Id(x =>
> >> x.Id).Column("GUID_PIPELINE_STEP_PARAMETER").GeneratedBy.Assigned();
>
> >> 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].
> >> 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.
>
> --
> Ramon

-- 
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.

Reply via email to