First tip: I’ve found recently that if I have a boolean field in the database that is being used to mark whether some event occurred (referrals.pending, or feeds.subscribed) it is often more effective to make the field a datetime and record the moment that the event occurred. Then, a NULL can be used to indicate that the event has not yet occurred. Thus, you have referrals.applied_at with a method on Referral like this:

1
2
3

def pending?
  applied_at.nil?
end

This gives you the capability down the road to not only report whether the event occurred, but how frequently over various periods of time.

The second tip is handy when you’re working on a migration. I find that the process (for me) works like this:

• Create the migration and run it.
• Discover I forgot something.
• Migrate down to the previous schema version.
• Change the migration and run it again.

(Repeat as necessary.) However, being the imperfect programmer that I am, I find that I often implement the #down method incorrectly, forgetting to drop a table or remove a column. Thus, when I try to run the migration again, it fails saying that the table/column already exists.

Using script/console and ActiveRecord::Schema, it becomes a cinch to clean up the artifacts:

1
2
3
4
5

ActiveRecord::Schema.define do
  drop_table :foos
  remove_column :bars, :blitz
  remove_column :bars, :things
end

And if you have the verbose_migrations plugin installed, you can also get some very useful output describing what was done and how long it all took.