Hi Nick (et al), I've probably mentioned it on the list before, but at last year's Zend/PHP Conference, I was quite impressed by a presentation by Mike Ho on the Qcodo framework (http://www.qcodo.com/).
This software has a code generator that you run against your database schema. The code generator introspects your database schema and generates PHP classes, even honoring foreign key relationships. The classes in which you write augmenting code extend the generated classes, so you don't lose this code if your database schema changes; you just need to rerun the code generator. In this way, the database introspection does not run on each request to the website, but only when you run the code generator. It also has a fairly robust forms package and is an good fit for RAD and prototyping. I just thought to mention it since you all were talking about ORM, and maybe we consider the ideas as we research solutions for the Zend Framework. Best regards, Darby Nick Lo wrote: > I've just recently had time to test out Zend_Db_Table and while I > realise it has a new proposal (which I can't get to right now as the > wiki appears to be down) I'm curious as to how many are really using or > planning to use it? > > I think the main thing that I don't feel comfortable with about its (and > ActiveRecord type) approach is the need to call DESCRIBE before it can > operate. This means that even the simplest query becomes at least 2 > queries and seems like an unnecessary overhead. I just wonder quite what > kind of apps are in need of this kind of thing. Small apps would surely > just be able to work more directly with the SQL and larger apps would > surely suffer performance issues. > > It would make sense to have some kind of cache of the database info > rather that be querying it afresh each time and only update the cache > when the database itself changed. > > Anyway, I am interested in any discussion about this particularly from > those who have a database admin leaning and have opinions about its > efficiency. > > Thanks, > > Nick > >
