Dear Bertrand and Stefane, to get the thread back on-topic, some in-line replies ;-)
Am 25.07.2011 um 11:38 schrieb Bertrand Delacretaz: > On Mon, Jul 25, 2011 at 11:20 AM, Stefane Fermigier <[email protected]> wrote: >> On Jul 25, 2011, at 10:44 AM, Bertrand Delacretaz wrote: >>> ...ORM is an anti pattern anyway ;-) > >> ...since Sebastian and his team made a great deal of effort to create their >> own ORM ("Object-RDF mapping") >> in the first place, we'd better acknowledge instead that "object-something >> mappings" are useful tools, >> with (as all tools) some limitations that one needs to be aware of... > > To me the main problem with SQL-to-object mappers is, as explained in > the seldo.com blog post, that if SQL is the wrong model for your data > (i.e. your data is not relational), mapping that to objects won't > help. I just creates more mess ;-) > > I haven't looked closely at what Sebastian's team have implemented in > this case, but an object to RDF mapping might make perfect sense if > RDF is the right model for your project and the mapped objects help > you access it. No impedance mismatch in that case. What we are really doing is an RDF-to-relational database mapping, and ORM is mainly the vehicle to create an abstraction from the different database vendors. It would not be so difficult to get rid of the ORM completely, because we are essentially just working with relational data. Performance- and complexity-wise it would definately make sense. But we would loose database independence, an argument that was often brought before, because companies typically already have a certain database landscape, they have DBAs that are expert on one system, they have fast servers running already one system, backups in place, ... Greetings, Sebastian -- | Dr. Sebastian Schaffert [email protected] | Salzburg Research Forschungsgesellschaft http://www.salzburgresearch.at | Head of Knowledge and Media Technologies Group +43 662 2288 423 | Jakob-Haringer Strasse 5/II | A-5020 Salzburg
