To clarify on my last post ORMs are designed only for records of relational data. Not all data is relational. Some is hierarchical and some may even be graph like. SQL can sometimes deal with this after a fashion (ie: self referencing table joins) but even then the results there are limited. Just not as limited as most ORMs. This is where I see the most significant impedance mismatching where ORM only handles a subset of SQL capability.
ORMs were only ever intended to map relational data to object models. New emergent database technologies like Mongo, Google App Engine and others may include non-relational data. Relational implies consistent data types across records. Non relational has no such constraint. Key-Value stores are one such example.
This also explains my comment about article posted in url link in previous post in this thread dated 2006 as being more relevant today. Their was an impedance mismatch back then (perhaps due to the relative immaturity of ORM technology back then and its failure to embrace the full power of SQL) and then is an impedance mismatch today (because database technology has moved on).
Maybe it will be solved one day with an “object non-relational mapper”. Who knows.
Sent from my XT316 using Tapatalk 2