Doing so would be harmless, except when the database does not (yet) exist. Note that running rake db:create loads AR, which in turn loads foreigner. Instead, pull it from the configured connection pool. The :adapter key here is the same as pulled from databases.yml for the current environment. This is slightly different from using #adapter_name which is a per-adapter defined constant that *incidentally* is the adapter name when downcased.
…abase connection is not available for some reason. This marks a new gem release with support for Rails 3.
…ad to only load Foreigner if ActiveRecord is loaded. This maintains fast environment startups, unless you have another crappy gem or plugin that causes lazy loading to begin.
Some folks wanted Oracle and Sql Server support, which currently required them to fork this project or copy/paste. That sucked, but at the same rate I only want to have adapters that Rails supports. Now just do: Foreigner.register 'oracle', /path/to/oracle/adapter
…le_name_suffix. Thanks accy.
…ditional foreign key
…. This code does not have context of the table name, which means that the plugin is unable to automatically generate a name for the foreign key. Since you can't use add_index inside a create_table block either, I think this is an OK removal to be consistent with existing migration APIs.
…postgres aren't consistent. Thanks bkaney
* Foreigner is now available on gemcutter. Add :source option to README
…g which adapter to use.
…ach do it differently. There's some duplication here but it's not terrible.
…he adapters to schema_dumper.
…heme and adding to schema dump