Larry Garfield was pointing out the mess would be dealing with different MySQL engines. My pointer to a benchmark of pgsql raised a lot of fuss. I'm developing a couple of modules that will run on pgsql and I noticed how much I relay on "on delete cascade" and how much pain it should be in terms of performances relying on "delete" users and node hooks when you mass delete stuff. MySQL already have those features, but I wouldn't dare to use them right now. Anyway they will become more mature and more available (as I said several times I like to have the chance to be unfaithful, and the fact that MySQL is becoming a better product for the things I need is much welcome). Before laying down any plan for DB abstraction in D7 it would be nice to see what are the common *easy to support* features across a bunch of Free/non-Free(?) DB that people use most. MySQL advocates consider this: the most used Open Source DB is going to offer reliable(?) data integrity features to the masses shortly(?). As other CMS will move to PHP5 and they will enjoy the benefits, they will start to use more advanced DB features and enjoy the features. As a side note I do really appreciate the fact that in MySQL you can chose between different engines, I consider it a configuration options and I do understand that MyISAM may be more suited for a "static" CMS and that people do care about performances, but again... I see Drupal as a CMS framework rather than a CMS so it should be suited for different tasks. There are tasks (and some of them are the tasks where the money are) which aren't suited to MySQL, at least now... -- Ivan Sergio Borgonovo http://www.webthatworks.it