Skip to content

Latest commit

 

History

History
108 lines (73 loc) · 3.69 KB

README.md

File metadata and controls

108 lines (73 loc) · 3.69 KB
Raw

activerecord-clean-db-structure

Ever been annoyed at a constantly changing db/structure.sql file when using ActiveRecord and Postgres?

Spent hours trying to decipher why that one team member keeps changing the file?

This library is here to help!

It cleans away all the unnecessary output in the file every time its updated automatically. This helps avoid merge conflicts, as well as increase readability.

Installation

Add the following to your Gemfile:

gem 'activerecord-clean-db-structure'

This will automatically hook the library into your rake db:migrate task.

Supported Rails versions

Whilst there is no reason this shouldn't work on earlier versions, this has only been tested on Rails 4.2 and newer.

It also assumes you use ActiveRecord with PostgreSQL - other ORMs or databases are not supported.

Caveats

Currently the library assumes all your id columns are either SERIAL, BIGSERIAL or uuid. It also assumes the id is the primary key.

Multi-column primary keys, as well as tables that don't have id as the primary key are not supported right now, and might lead to wrong output.

You can disable this part of the cleaning process in your config/environments/<environment>.rb (or config/application.rb):

Rails.application.configure do
  config.activerecord_clean_db_structure.ignore_ids = true
end

Other options

You can optionally have indexes following the respective tables setting indexes_after_tables:

Rails.application.configure do
  config.activerecord_clean_db_structure.indexes_after_tables = true
end

When it is enabled the structure looks like this:

CREATE TABLE public.users (
    id SERIAL PRIMARY KEY,
    tenant_id integer,
    email text NOT NULL
);

CREATE INDEX index_users_on_tentant_id ON public.users USING btree (tenant_id);
CREATE UNIQUE INDEX index_users_on_email ON public.users USING btree (email);

To enable sorting the table column definitions alphabetically, discarding the actual order provided by pg_dump, set order_column_definitions:

Rails.application.configure do
  config.activerecord_clean_db_structure.order_column_definitions = true
end

You can have the schema_migrations values reorganized to prevent merge conflicts by setting order_schema_migrations_values:

Rails.application.configure do
  config.activerecord_clean_db_structure.order_schema_migrations_values = true
end

When it is enabled the values are ordered chronological and the semicolon is placed on a separate line:

INSERT INTO "schema_migrations" (version) VALUES
 ('20190503120501')
,('20190508123941')
,('20190508132644')
;

By default the gem will remove some extensions that typically aren't needed in structure dumps. You can choose to keep all, or just some, of those extensions:

Rails.application.configure do
  config.activerecord_clean_db_structure.keep_extensions = :all

  # This does the same thing as :all. You can choose which optional extensions to keep.
  config.activerecord_clean_db_structure.keep_extensions = ["pg_stat_statements", "pg_buffercache"]
end

Authors

License

Copyright (c) 2017, Lukas Fittl
activerecord-clean-db-structure is licensed under the 3-clause BSD license, see LICENSE file for details.