I've also spent a lot of time looking through the sample applications trying to find out how JPA works with Aries. In my case I use managed JPA which I think is closer to how the samples work. I agree with Harald that this is pretty tough to grasp without clear and concise documentation. It would be extremely useful with a tutorial (not a sample) where all the necessary steps are described.
BTW I have been using "aries:" and not "osgi:" when referring to a published service and I did not know that Aries JNDI was required for that. I thought a namespace handler for this namespace was part of "core" Aries Blueprint. Good evidence that better documentation is really needed. /Bengt 2010/9/17 Alasdair Nottingham <[email protected]> > On 16 September 2010 11:31, Harald Wellmann <[email protected]> > wrote: > > > > Another question: how does the Persistence Provider discover the data > source - where does the magic happen so that the lookup of > <jta-data-source>osgi:service:/javax.sql.DataSource</jta-data-source> will > work? Is that done by Aries alone, or does the persistence provider need to > be OSGi aware in this respect? > > > > The osgi:service/javax.sql.DataSource (your example had an extra : I > removed), is a JNDI lookup into the service registry. Aries does not > have any code for creating DataSources, but the blog and aries trader > sample use blueprint to cause a derby datasource to be defined and > registered, I suspect you can use DS or anything to get one > registered. You then need the aries JNDI sub project which makes this > all work. > > > Thanks in advance for any hints! > > > > Best regards, > > Harald > > > > > > -- > Alasdair Nottingham > [email protected] >
