Mailing-List: contact users-help@jackrabbit.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@jackrabbit.apache.org
Received-SPF: pass (nike.apache.org: domain of lists@nabble.com designates
 216.139.236.158 as permitted sender)
Message-ID: <24699832.post@talk.nabble.com>
Date: Tue, 28 Jul 2009 07:51:02 -0700 (PDT)
From: Gadbury <gadbury@googlemail.com>
To: users@jackrabbit.apache.org
Subject: Re: User Managed Transactions...
In-Reply-To: <24687924.post@talk.nabble.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
References: <24687924.post@talk.nabble.com>


Sorry to bump the post but I thought I should clarify my question :)

If I am within a user managed transaction's scope, and I have added a node
and then call session.save(), is it safe to rely on that node's UUID or
would I have to wait until the transaction has been committed?

Here is the excerpt from the JCR 1.0 specification:


=094.9.1.1 When UUIDs are Assigned

=09In some client-server implementations the assignment of a permanent UUID
may be done on the server.=20
=09In these cases it is not practical for a newly created referenceable nod=
e
to be given a UUID upon creation.=20
=09Rather, it makes more sense for the UUID to be assigned upon save of tha=
t
node.=20
=09In such cases a =E2=80=9Cdummy UUID=E2=80=9D may be assigned on creation=
 of a new node
while the real UUID assignment takes place later, upon save.
=09Applications should not, therefore, rely on the UUID of a node before th=
at
node is saved for the first time.=20


Thanks again.  Kind regards,

James


Gadbury wrote:
>=20
> Hi all,
>=20
> I have a number of Jcr-based Daos.  In some cases, I have been using the
> session as a means to control a simple transaction.  For example, I creat=
e
> a session and pass it to two different Dao create() methods.  Only when
> all my Dao work is complete do I call session.save().  See my example cod=
e
> below:
>=20
>=20
> =09Session session =3D getSession();
>=20
> =09JcrUserDao userDao =3D new JcrUserDao();
> =09JcrProductDao productDao =3D new JcrProductDao();
>=20
> =09userDao.create(aUser, session);
>=20
> =09productDao.create(aProduct, session);
>=20
> =09session.save();
>=20
>=20
> The problem I have is that during a single transaction, I would like to
> create a multiple value property that contains strings for all node UUIDs
> created during the single transaction (i.e. hold 'weak' references to any
> newly created node UUIDs).  I read in the specification that it is not
> safe to use a referenceable node's UUID until session.save() has been
> called and the changes have been persisted.  However, with the above
> example, if I call save() before the two create methods have completed, I
> ruin my simple transaction.
>=20
> I have been reading the spec again regarding user managed transactions.=
=20
> Consider the following code:
>=20
>=20
> =09Session session =3D getSession();
>=20
> =09// Get user transaction (for example, through JNDI)
> =09UserTransaction utx =3D ...=20
>=20
> =09JcrUserDao userDao =3D new JcrUserDao();
> =09JcrProductDao productDao =3D new JcrProductDao();
>=20
> =09utx.begin();
>=20
> =09Node newUserNode =3D userDao.create(aUser, session);
>=20
> =09session.save();
>=20
>         String UUID =3D newUserNode.getUUID();
>=20
> =09Node newProductNode =3D productDao.create(aProduct, session);
>=20
> =09session.save();
>=20
> =09utx.commit();
>=20
>=20
> After the first call to session.save(), would it then be safe to get a
> node's UUID?
>=20
> For example, in my JcrProductDao's create() method, I am creating one
> Product node and then create N number of ProductElement nodes by calling =
a
> JcrProductElementDao, again passing on the session.  I would like a
> multiple value property of a Product to hold all new Element node UUIDs.
>=20
> However, I also read this in the JCR 1.0 spec:
>=20
> =09"Note, however, that changes made in the transient storage are not
> recorded by a transaction.
> =09This means that a rollback will not revert changes made to the transie=
nt
> storage of the Session.
> =09After a rollback the Session object state will still contain any pendi=
ng
> changes that were present before the rollback."
>=20
> Does this mean that I shouldn't be using session and making transient
> changes with user managed transactions?  I don't mind if I cannot revert =
a
> session's pending changes just as long as I can control when the changes
> are persisted and also reliably access a new node's UUID.
>=20
> Any help or advice would be greatly appreciated.  Thanks for reading.
>=20

--=20
View this message in context: http://www.nabble.com/User-Managed-Transactio=
ns...-tp24687924p24699832.html
Sent from the Jackrabbit - Users mailing list archive at Nabble.com.