Jonathan Chaffer wrote:
A simpler implementation is to load each field through a separate query. This is a very flexible approach, and allows simple compartmentalization of the query code into the field modules. The obvious downside is that you need many queries for a simple node load. 1x query to fetch node IDs for the ten nodes (10 results) 100x queries to fetch each normal field (1 result each) 10x queries to fetch the multi-value fields (10 results each)
111 queries is not acceptable. Using a node cache is essential. Where to store the cached nodes? Well, I'd expect the answer to depend on how we are going to deal with node revisions. Plus, no decision is set in stone. We can move the cached copies around if necessary. -- Dries Buytaert :: http://www.buytaert.net/