Kristian is currently talking about the new NdbRecord API for the NDBAPI and how it relates to ha_ndbcluster (the mysql storage engine, which uses ndbapi to talk to the cluster nodes) and how it can be used by ndbapi applications.

It looks like we’re getting a really neat API that avoids so much mess and makes it possible to write incredibly efficient mappings between what comes over the wire from data nodes and whatever internal structures the application wants to fill out.

Talking about this and Monty Taylor’s ORM mapping stuff could be very interesting.