Using fixtures is unwieldy to work with with models with many relations. Many tests depend on the same data scattered among many fixtures. This makes it difficult to add new data for new tests as things like counting the number of particular types of records change when I add new records for the new tests. It just really loses that clean Ruby/Rails feeling. I''m thinking it might be better to build my test data in code in the setup methods. That keeps it all in one place and independent of other test sets. Anyone want to share thoughts or experiences about working that way? -- Jack Christensen jackc-/SOt/BrQZzOj3I+7jmQ39gC/G2K4zDHf@public.gmane.org