MigLib 3.0.0

Migrate is a SQLite-first migration toolchain built around F# schema scripts (.fsx). It provides a hot-migration workflow (migrate → drain → cutover) plus type-safe code generation from reflected schema types.

Installation

If you just want to test the tool without installing .Net, then you can use a Docker image:

podman run -it 'mcr.microsoft.com/dotnet/sdk:10.0' bash

Inside the container run:

export PATH="$PATH:/root/.dotnet/tools"

After having .Net in your system you can run

dotnet tool install --global migtool

Local Tool Build/Install (FAKE)

Build, pack, and install the current branch as a global mig tool from the local package output:

dotnet fsi build.fsx

Run specific FAKE targets when needed:

dotnet fsi build.fsx -- --target Build
dotnet fsi build.fsx -- --target PackTool

Run the installed local tool directly:

mig --help

Quickstart (Online Migration)

Assuming:

• an existing SQLite database named <dir>-<old-hash>.sqlite
• a target schema script at schema.fsx

# from your project directory:
# - expects ./schema.fsx
# - expects exactly one source db matching <dir>-<old-hash>.sqlite
# - derives target db as <dir>-<schema-hash>.sqlite
mig plan
mig migrate

# then continue in the same directory (paths auto-resolve)
mig status
mig drain
mig cutover
# optional, after traffic has fully moved to the new service:
mig cleanup-old

# from a different directory:
mig migrate -d /path/to/project

# if migrate fails and you need to clear failed migration artifacts:
mig reset --dry-run
mig reset

Quickstart (Schema Initialization)

When you want to bootstrap a database directly from schema.fsx (no source DB yet):

# from your project directory:
# - expects ./schema.fsx
# - derives db path as <dir>-<schema-hash>.sqlite
mig init

Features

• F# schema reflection from .fsx scripts
• FK-aware bulk copy and replay with ID mapping
• Replay checkpoints and drain safety validation
• Schema identity metadata (schema_hash, optional schema_commit) persisted in new database
• Operational status reporting for old/new database migration state
• Optional old-database migration-table cleanup after cutover

Specs

• specs/database_dsl.md
• specs/hot_migrations.md
• specs/mig_command.md
• specs/operator_runbook.md

Commands

• mig init [--dir|-d <path>] - Create a schema-matched database from schema.fsx and apply seed inserts (no source DB required).
• mig codegen [--dir|-d <path>] [--module|-m <name>] [--output|-o <file>] - Generate F# query helpers from schema.fsx into a file in the same directory as the schema.
• mig migrate [--dir|-d <path>] - Create the new DB from schema, copy data, and start recording on old DB.
• mig plan [--dir|-d <path>] - Print dry-run inferred paths, schema diff summary, and replay prerequisites without mutating DBs.
• mig drain [--dir|-d <path>] - Switch old DB to draining mode and replay pending migration log entries.
• mig cutover [--dir|-d <path>] - Verify drain completion plus old marker/log replay safety, switch new DB to ready, and remove replay-only tables.
• mig cleanup-old [--dir|-d <path>] - Optional cleanup of old DB migration tables (_migration_marker, _migration_log).
• mig reset [--dir|-d <path>] [--dry-run] - Reset failed/aborted migration artifacts, or inspect reset impact without mutating DB files.
• mig status [--dir|-d <path>] - Show marker/status state and migration counters for operational visibility.

Contributing

How to contribute:

• Open an issue to discuss the change and approach
• Add relevant tests
• Create a pull request mentioning the issue and also including a summary of the problem and approach to solve it
• Wait for the review

Acknowledgments

This project wouldn't have been possible without the amazing open-source community. We're especially grateful to:

• Fabulous.AST - An elegant F# DSL for code generation that made creating and manipulating F# AST a joy
• Fantomas - The excellent F# code formatter that ensures our generated code is beautiful and consistent

If you find these projects valuable, please consider supporting them:

• Star their repositories
• Contribute to their projects
• Donate to support their continued development

License

Apache 2.0