crimson/build/db-init

db-init

This script setups of the databse with the requrired baseline, runs migrations, and loads the api schema (if enabled).

Migration script

All migrations scrips MUST do ALL of the following:

- Placed in src/db/migrations
- Named with its migration number (0 indexed), and have four
  numbers of padding. i.e. `0000.sql`, `0030.sql`, or `9999.sql`
- In numerical order with all other migrations (cannot go from migration
  0 to 2).
- A postgres transaction. `BEGIN TRANSACTION ... COMMIT TRANSACTION`.
- End with the following before COMMIT, where <rev> is the NEXT
  revision number. (i.e. in `0000.sql` <rev> MUST be 1).

UPDATE sys.database_info SET curr_revision = <rev> WHERE name = current_database();

Example 0000.sql:

BEGIN TRANSACTION;

CREATE SCHEMA website;

UPDATE sys.database_info SET curr_revision = 1 WHERE name = current_database();

COMMIT TRANSACTION;

Migrations will ONLY EVER be ONCE, and will ALLWAYS be run in order. This means that you can assume all previous migrations have run successfully in any current migration, and there is NO other possible state in the database.

API

Once all migrations have been completed, the api will be initalized (if enabled.

If you opt to use postgrest which is builtin into crimson, you must create a sql file loads the api schema that MUST do ALL of the following:

- Placed in src/db/rest and named `rest.sql`
- A postgres transaction. `BEGIN TRANSACTION ... COMMIT TRANSACTION`.

Within that transaction you can setup postgres with the api schema you want. See https://docs.postgrest.org/en/v12/. (crimson currently uses postgres 12).

NOTE: If you want to load any sql file though an absolute path, src/db will be mounted as READ ONLY to /db. (i.e. src/db/rest/rest.sql => /db/rest/rest.sql).