Test for eager associations with limits should not assume that records are ordered by id.
r2737@asus: jeremy | 2005-10-24 19:06:09 -0700
Fail fast if invalid primary key column.
r2746@asus: jeremy | 2005-10-25 15:37:28 -0700
Begin rollback of fixture delete order. Its solves a problem for 1% of users who already have a workaround while severely slowing down the other 99%.
r2747@asus: jeremy | 2005-10-25 16:03:01 -0700
Rollback the rest.
r2748@asus: jeremy | 2005-10-25 16:06:26 -0700
Re-add fixtures declaration to conditions scoping test
r2749@asus: jeremy | 2005-10-25 16:09:03 -0700
Re-add fixtures declaration to finder test
r2750@asus: jeremy | 2005-10-25 16:13:50 -0700
Don't assume keyboards table is empty
git-svn-id: http://svn-commit.rubyonrails.org/rails/trunk@2730 5ecf4fe2-1ee6-0310-87b1-e25e094e27de
Split association reflection test into has_many and has_one tests. TODO: extend to belongs_to and habtm as well. Kill trailing whitespace.
r2980@asus: jeremy | 2005-07-10 01:53:41 -0700
Test that eager load of firm.account (a has_one) returns the correct account. This doesn't cover the regression in Ticket #1688, however. Need a case where class name is not provided and the singular name is over-singularized by active_record.class_name. Kill trailing whitespace.
r2981@asus: jeremy | 2005-07-10 01:56:44 -0700
Fix class name resolution in AssociationReflection#name_to_class_name. Use the equivalent of the idiomatic ||= (read or write if nil) for the class_inheritable_attribute reflection arrays. Kill trailing whitespace.
r2982@asus: jeremy | 2005-07-10 01:59:08 -0700
Remove superfluous assignment in compute_type. In an inject, the return value is used for the next injection, so assigning to the injected value is unnecessary.
git-svn-id: http://svn-commit.rubyonrails.org/rails/trunk@1790 5ecf4fe2-1ee6-0310-87b1-e25e094e27de
Jeremy Kemper