I have several tables that need to be populated when I move my project to production, each of these tables has several thousand rows. I have the data stored in a CSV file now, but using the seed.rb file seems like it would be cumbersome because the data from my CSV file would have to be formatted to meet the seed.rb format. If this were only a handful of rows, it wouldn't such a problem. What would be the best/easiest way to get this data loaded?
I would probably use a little custom script and the faster_csv gem which has good tools to parse .csv files quickly. Then you can map the fields to model attributes.
I would implement this via TDD as model methods and use ActiveRecords's create method to instantiate instances. While this is slower than writing SQL straight out, it's safer in that your data will run through all the model validations and you have a better confidence in the data integrity.
Flushing out data integrity issues from legacy data import up front will save you a lot of trouble later.