On 2010-01-29 04.14, Niclas Hedhman wrote:
* Tables exists and populated. The structure was defined by a
previous application or team, and will not change.
Meaning, it is outside the scope of Qi4j to map an arbitrary domain
model to an RDBMS, and only pull in an existing schema with all its
ugliness, try to hide it and re-exposed via specific code for it.
This can probably be done in many ways, but I don't think that an
arbitrary domain model in Qi4j should be possible to map to an
arbitrary SQL schema. After all, it is possible for programmers to
do additional 'mapping' in Qi4j-to-Qi4j code.
Would this map to my 1.1 case in the previous post? I.e. tables exist,
but we want to do a read/write app for it, hence doing a one-table
solution with a mapping from that to whatever existed before is ok?
As for "search for all HasName", RDBMS support doesn't mean that we
will try to do queries against the raw data via the Fluent Query
API. I think that is a loosing battle not worth fighting. However,
named native queries could be used for direct SQL, and the developer
is in control.
I agree on that. If we use the one-table solution you'd still have the
RDF-repositories or Lucene for the querying, and then named queries if
you want to do SQL.
/Rickard
_______________________________________________
qi4j-dev mailing list
[email protected]
http://lists.ops4j.org/mailman/listinfo/qi4j-dev
