On Thu, 2013-02-07 at 15:35 +0100, mfoj...@redhat.com wrote: > From: Michal Fojtik <mfoj...@redhat.com> > > Currently everytime we add or remove attributes from > DB, users need to rerun the 'rake' task to recreate the > database. > > This patch will make possible to update the DB schema > without executing the rake task. > > The 'migrations' extension in Sequel will check the 'schema_version' > table and apply all migrations >= than the current schema. > > Migrations must be prefixed with the schema number in chronological > order and they are stored in server/lib/db/migrations.
We definitely need migrations - I don't like though that they are automatically applied when the server starts. They should be applied by a separate command (either a rake task or a small script in bin/) - this is important both for dev, so you can go back and forth between the schema changes you're working on, and for prod where you want to make sure people take a DB dump etc. before migrating the database. Also, migrations should go into server/db/migrations, since they are not needed (or wanted) while the server is running. David
