Phoenix Less Seeds

Reducing the burder of seeds.exs scripts #

At some point the migration files become so numerous that it is preferable to consolidate them into a structure.sql file so that they can be loaded using mix ecto.load. This is already well documented in this post from fly.io Developing after mix ecto.dump. But what about the seeds.exs scripts?

In my current project there is a large set of what we refer to as “lookup” tables, they could also be referred to as a data dictionary. All the data shown in dropdown lists etc. comes from these lookup tables and they are (by convention in this project) prefixed with lk_ and usually just contain an ID and a name. These “lookup” tables are populated by seeds.exs scripts. Over time the number and complexity of these scripts increased and they need to be maintained as items are re-ordered or certain items are no longer needed etc.

In order to reduce the maintenance burden all the seeds files were consolidated into the second part of structure.sql. The following shows the commands that were used to create the sql statements.

sudo -iu postgres
pg_dump --dbname=myapp_dev --file=myapp-dev.sql --table="lk_*" --data-only --inserts

That command can be run on a local dev machine against a clean database. The contents of the produced file myapp-dev.sql is then append to the structure.sql file.

And that’s it - short and sweet but it gives us a new base to work from and less code to maintain - happy days!