Organizing Migrations

[scal-lider]

Hi there.
I have too many migrations T_T, so I would like to organize it in folders.
It is possible?
As far as I know, the migration runs in alphabetical order, if it is possibles, the folder needs to be in alphabetical order too? Or only the files inside folders?

In the example below, wich migration runs first?

a/
|-- 1_populate_table
b/
|-- 0_create_table

[e-e]

This is pretty easy to test - just try it out:

With a folder structure like:

database/
|-- migrations/
|---- 00/
|------ 1618549676073_test_3_s
|------ 1618549678710_test_4_s
|---- 01/
|------ 1618549669282_test_1_s
|------ 1618549672676_test_2_s

it ran in the following order:

+ node ace migration:run
❯ migrated database/migrations/00/1618549676073_test_3_s
❯ migrated database/migrations/00/1618549678710_test_4_s
❯ migrated database/migrations/01/1618549669282_test_1_s
❯ migrated database/migrations/01/1618549672676_test_2_s

so yes, it will run all migrations in folder "A" before folder "B"

[RomainLanz]

Hey @scal-lider! 👋

We are using a deep natural sort.

In other terms, your folders and files are run in your expected order. 👍

[scal-lider]

So I have to search "natural sort" 😅
Basically it transforms strings in tuples with strings and number and compare them, so:

"test_10" > "test_2"  is true

"test_10" => ["test_", 10]
"test_2" => ["test_", 2]

"test_" == "test_" AND 10 > 2

When compare a string with a number, it uses a string comparation:
'a' > '1618549676073'

Based on the @e-e's test, the sort also count the folder's name:
00/1618549676073_test_3_s < 01/1618549669282_test_1_s,
And probably
a/0_test > 1618549676073_test > 00/1618549676073_test_3_s
since 'a' > '1' and 'a' >'0' and 1618549676073 > 0

[Xqua]

Hey! I just built a tool that might help with this: adonis-lucid-migration-squash

Instead of organizing migrations into folders, you can consolidate/squash them into a single baseline migration (similar to Django's squashmigrations).

What it does:

• Takes your production database schema
• Generates a clean, single Knex migration file
• Includes automated verification to ensure 100% accuracy
• Reduces deployment time (1 migration instead of 100+)

Quick example:

# Dump your schema
pg_dump -s --no-owner --no-acl > schema.sql

# Convert to Knex migration
python -m pg_to_knex schema.sql baseline_migration.ts

# Done! Clean baseline migration with verification