Diego, I fear you've misunderstood what I'm trying to do.  I'm not
trying to protect myself form the compiler, but to have a semantically
complete API.  I don't simply pass around person names as 'strings'
but as PersonName's, which have a stringer semantic to them than just
a string (as well as constraining their possible domain of values).
I've been through many projects on many jobs where problems arise from
anemic types.

I can use strong, semantic types for any general property in an entity
destined for mapping in NHibernate by using <composite>.  However, I
can't use that for an ID.

I see that you view these "wrappers" as nonsense, but I argue they're
far from it.

As for being easier to work with the frameworks, I've found no
friction from doing it this way *except* for identifiers.

Regards,
Brian.

On Jun 14, 2:46 pm, Diego Mijelshon <[email protected]> wrote:
> So, you essentialy tried to protect you from yourself by abusing the
> compiler.
> Why stop with IDs? Write a wrapper for every property, so you can't
> accidentally assign a ClientBirthDayDateTime to a PaymentDueDateDateTime...
>
> My suggestion: remove all that wrapper nonsense. It's easier to work with
> the frameworks than against them.
>
> For the record, "var" doesn't have a type, it's a keyword that triggers
> implicit typing.
>
>    Diego
>
>
>
> On Mon, Jun 14, 2010 at 11:04, Trinition <[email protected]> wrote:
> > Sure, an int is more strongly typed than 'var', but one can't
> > distinguish an int representing a Client's ID from an int representing
> > an Organization's ID.  But surely you wouldn't want to confuse the
> > two.
>
> > So far, I have been pointing to a private field.  The problem occurs
> > with NHIbernate's LINQ because the public property I'm exposing to my
> > domain (including my repositories which use LINQ) aren't known to be
> > mapped by NHibernate, so it can't convert them to a query.  For
> > example:
>
> > class Client {
> >  private int _id; // Mapped as the identity in Client.hbm.xml
> >  public ClientId Id { get { return new ClientId(_id); } } // A
> > generated, strongly-typed wrapper
> > }
> > class ClientRepository {
> >  public IList<Client> GetSpecificClients(IList<ClientId> clientIds)
> > { // This is a contrived example
> >    return _unitOfWork.Find<Client>().Where(client =>
> > clientIds.Contains(client.Id)).ToList(); // 'Id' property of Client is
> > not known to NHibernate!
> >  }
> > }
>
> > That is why I'd prefer to use my ClientId class directly as the
> > identity type for Client.
>
> > Regards,
> > Brian.
>
> > On Jun 11, 9:04 pm, Diego Mijelshon <[email protected]> wrote:
> > > How is an int _not_ strongly-typed? What value do those wrappers provide?
> > > In any case, you can always point NHibernate to your private field, and
> > it
> > > will skip your wrapper completely:
>
> > > <id name="_Id" access="field" type="Int32">
> > >   <generator class="..."/>
> > > </id>
>
> > >    Diego
>
> > > On Fri, Jun 11, 2010 at 16:12, Trinition <[email protected]> wrote:
> > > > In my domain, I'd really prefer to have strongly-typed IDs (simple
> > > > immutable wrappers around ints).  This would allow me to have clear
> > > > method signatures and allow Intellisense to assist me more accurately:
>
> > > > public void ResetFirmLockout(FirmId firmId); // Consider this strongly-
> > > > typed identity...
> > > > public void ResetFirmLockout(intfirmId); // Consider this weakly-typed
> > > > identity...
>
> > > > It may sound trivial, but I've pulled it off half-way and it has been
> > > > quite nice to have it this way (I'm using a private int field "_id"
> > > > but exposing a readonly-property to the domain of type FirmId that is
> > > > constructed on the fly).
>
> > > > The only problem is that I can't get NHibernate to recognize a
> > > > component as it's identiy.  The "<id...>" takes a Type that must be an
> > > > NHibernate type like "Int32".  I can't specify a user type such as
> > > > "FirmId".
>
> > > > I've also considered <composite-id>, but it only works with
> > > > application-assigned identifiers, whereas I do want to rely upon DB-
> > > > generated identities.
>
> > > > I considered a custom IIdentityGenerator, but as best I can tell, it
> > > > would still have to be an NHibernate type (e.g. "Int32").
>
> > > > Has anyone else ever accomplished this?
>
> > > > --
> > > > 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%252bunsubscr...@googlegroup 
> > s.com>>
> > > > .
> > > > 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.

Reply via email to