find and execute trove migrations

trove_migrations.rb

# load all migrations
Dir[Rails.root.join('db/migrate/*.rb')].each { |f| require f }

#find the trove migrations classes
# and the corresponding file
# sort by file
# instantiate
# and invoke change
# rescuing from any exception.

trove_migrations = Trove::BaseMigration.subclasses.collect { |c| 
  [c] + Dir[Rails.root.join("db/migrate/*#{c.name.tableize.singularize.gsub('trofe','trove')}*.rb")]
}.sort{ |a, b| a[1] <=> b[1] }.collect{ |pair| pair[0] }

trove_migrations.each { |c|
  begin
    mig = c.new
    if mig.respond_to? :change
      mig.change
    else
      mig.up
    end
  rescue
  end
}

# TODO dump schema to trove_schema.rb

1. Do we have a file named trofe?
2. Some of the migrations call up, not change.
3. I'd make pretty sure I don't rescue from non-specific exceptions, otherwise you'd never know that the migration calls up instead of change, or that the migration failed because you already have the table, etc.
4. Couldn't you just load only the trove migrations?

1. trofe is the singular of trove!
2. yes I spotted that
3. true but I'm lazy...
4. we need to load them all to find which are subclasses of Trove::BaseMigration. However we could adopt the rule that all and only all trove migrations started (or ended) with trove.

Next step would be to dump the schema,
then at start of tests create the trove schema from trove_schema.rb