migra/docs/index.md
Robert Lechte 530a188198 doc tweaks
2020-08-20 19:28:20 +10:00

3.3 KiB

migra is a schema comparison tool for PostgreSQL.

It's a command line tool, and Python library. Find differences in database schemas as easily as running a diff on two text files.

migra makes schema changes almost automatic. Management of database migration deployments becomes much easier, faster, and more reliable.

Using migra is as simple as this:

$ migra postgresql:///a postgresql:///b
alter table "public"."book" add column "author" character varying not null;

alter table "public"."book" alter column "name" set not null;

To get started, hit the Quickstart guide.

Migra is 100% open source software. The code is available on github.

Features and Limitations

The following features of postgres are supported:

Feature Supported Notes/limitations
tables
partitioned tables
constraints
views
functions All languages except C/INTERNAL
indexes
sequences Does not track sequence numbers
schemas
extensions
enums
privileges Not exhaustive. Requires --with-privileges flag
row-level security NEW! Doesn't include role management
triggers
identity columns
generated columns
custom types/domains Basic support (drop-and-create only, no alter)

migra plays nicely with extensions. Schema contents belonging to extensions will be ignored and left to the extension to manage.

migra plays nicely with view/function dependencies, and will drop/create them in the correct order.

Endorsements

migra was used to manage the schema that powers PyPI:

Migra is cool as hell though, when I was trying to reconcile PyPI with the alembic initial schema I had to download some weird java app I found on some sketchy website to do that :P

I can definitely see Migra is more productive when switching around between schemas in development.

Development

Migra is developed on github. Contributions are welcome, get involved!

Philosophy

There was a good comment on Hacker News discussing how migra differs from traditional migration tools.

This is awesome!

A lot of people point to migrations as the best way to track changes to a database schema. But there are a lot of problems with them. For example, they involve adding version control on top of another version control system, which can cause a ton of problems. They also don't maintain themselves well. If you leave it alone, running migrations will just take longer and longer over time, even though you don't really get more utility.

I think we need more support from databases themselves to solve this problem.

In the meantime, this is a really good stopgap, because it can theoretically allow you to just have a file with your "ideal schema" for each commit. No need to maintain a separate database migration history too. You can even generate a series of migrations by looking at your git history!"

  • blaisio on Hacker News