persistent-migration: Manual migrations for the persistent library

[ bsd3, database, library ] [ Propose Tags ]
Versions [RSS] 0.0.1, 0.0.2, 0.1.0, 0.2.0, 0.2.1, 0.3.0
Change log CHANGELOG.md
Dependencies base (>=4.7 && <5), containers, fgl, mtl, persistent, text, time, unordered-containers [details]
License BSD-3-Clause
Author Brandon Chinn <brandonchinn178@gmail.com>
Maintainer Brandon Chinn <brandonchinn178@gmail.com>
Category Database
Home page https://github.com/brandonchinn178/persistent-migration#readme
Bug tracker https://github.com/brandonchinn178/persistent-migration/issues
Source repo head: git clone https://github.com/brandonchinn178/persistent-migration
Uploaded by brandonchinn178 at 2021-09-29T02:16:12Z
Distributions NixOS:0.3.0
Reverse Dependencies 2 direct, 0 indirect [details]
Downloads 2440 total (16 in the last 30 days)
Rating (no votes yet) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Docs available [build log]
Last success reported on 2021-09-29 [all 1 reports]

Readme for persistent-migration-0.3.0

[back to package description]

persistent-migration

CircleCI Hackage

This is a migration library for the persistent package.

Overview

By default, persistent provides a way to do automatic migrations; how to quickly and conveniently update the schema to match the definitions of the models. Because of its automatic nature, it will also balk at any operations that may delete data ("unsafe" migrations).

However, in a lot of production cases, you don't actually want this automatic migration. You might want to be able to run certain unsafe migrations because you know a column is safe to delete. You might want to be able to copy and transform data from one column to another and then delete the old column. You might want explicit/manual migrations for other reasons.

This package exposes an Operation data type that will be converted into SQL by a persistent backend. To define a series of migrations, write a list of these Operations and call runMigration from the appropriate backend module. Each Operation represents a movement from one version of the schema to another. runMigration will check to see the current version of the schema and run the Operations necessary to get from the current version to the latest version.

Usage

import Database.Persist.Migration

createPerson :: CreateTable
createPerson = CreateTable
  { name = "person"
  , schema =
      [ Column "id" SqlInt32 [NotNull, AutoIncrement]
      , Column "name" SqlString [NotNull]
      , Column "age" SqlInt32 [NotNull]
      , Column "alive" SqlBool [NotNull]
      , Column "hometown" SqlInt64 [References ("cities", "id")]
      ]
  , constraints =
      [ PrimaryKey ["id"]
      , Unique "person_identifier" ["name", "age", "hometown"]
      ]
  }

migrateHeight :: RawOperation
migrateHeight = RawOperation "Separate height into height_feet, height_inches" $
  map migrateHeight' <$> rawSql "SELECT id, height FROM person" []
  where
    migrateHeight' (Single id', Single height) =
      let (feet, inches) = quotRem height 12
      in MigrateSql "UPDATE person SET height_feet = ?, height_inches = ? WHERE id = ?"
        [ PersistInt64 feet
        , PersistInt64 inches
        , PersistInt64 id'
        ]

migration :: Migration
migration =
  -- first migration path should create all the tables
  [ 0 ~> 1 := [createPerson]

  -- can define shorter migration paths for equivalent operations; version 2, in this case, should result
  -- in the same schema, regardless of the path taken to get there.
  , 1 ~> 2 := [DropColumn ("person", "alive")]
  , 0 ~> 2 :=
    [ createPerson{ctSchema = filter ((/= "alive") . colName) $ ctSchema createPerson}
    ]

  -- example for adding columns
  , 2 ~> 3 :=
    [ AddColumn "person" (Column "gender" SqlString []) Nothing
      -- Adding a non-null column needs a default for existing rows.
    , AddColumn "person" (Column "height" SqlInt32 [NotNull]) (Just $ PersistInt64 0)
    ]

  -- example for more complex migrations; here, we split up the height field into feet and inches fields
  , 3 ~> 4 :=
    [ AddColumn "person" (Column "height_feet" SqlInt32 []) (Just $ PersistInt64 0)
    , AddColumn "person" (Column "height_inches" SqlInt32 []) (Just $ PersistInt64 0)
    , migrateHeight
    , DropColumn ("person", "height")
    ]
  ]
import Database.Persist.Migration (checkMigration, defaultSettings)
import Database.Persist.Migration.Postgres (runMigration)

-- the migration defined above
import MyMigration (migration)

-- the migration from persistent's mkMigrate
import MyMigration.Migrate (migrationDef)

main = do
  -- run the usual migration
  runMigration defaultSettings migration

  -- fails if persistent detects more migrations not accounted for
  checkMigration migrationDef

For more examples, see test/integration/Migration.hs.

FAQs

  • I don't know the SqlType corresponding to my column's Haskell type

As a general rule, complicated JSON serialization will be SqlBlob, but it might be SqlString for simpler data types. You can always choose one, and see if Persistent complains about it needing to be another type.

For example, you might want to put SqlBlob first, and see if Persistent errors with something like:

More migrations detected:
 * ALTER TABLE table ALTER COLUMN col TYPE VARCHAR

If Persistent tries to change the type to VARCHAR, then it probably wants SqlString instead.