Summary

Adds a complete migration system to Toasty: a CLI that diffs your Rust models against the last known schema snapshot, generates SQL migration files, and applies them to your database. Supports SQLite, PostgreSQL, and MySQL, with database-aware DDL generation (e.g. SQLite's table-recreation pattern for column type changes).

Usage

1. Configure your project with a Toasty.toml at the root:

[migration]
path = "toasty"
prefix_style = "Sequential"
checksums = false
statement_breakpoints = true

2. Wire up the CLI in a binary (src/bin/cli.rs):

use toasty_cli::{Config, ToastyCli};

#[tokio::main]
async fn main() -> anyhow::Result<()> {
    let config = Config::load()?;
    let db = create_db().await?;
    let cli = ToastyCli::with_config(db, config);
    cli.parse_and_run().await
}

3. Generate a migration when your models change:

cargo run --bin cli -- migration generate --name initial

This diffs your current schema against the last snapshot and produces a SQL migration file. If it detects a potential rename (table, column, or index), it interactively prompts you:

  Table "users" is missing
  > Drop "users" ✖
    Rename "users" → "people"

4. Apply pending migrations to your database:

cargo run --bin cli -- migration apply

5. Drop a migration (interactive picker or by name):

cargo run --bin cli -- migration drop --latest
cargo run --bin cli -- migration drop --name 0001_migration.sql

6. Print the current schema snapshot:

cargo run --bin cli -- migration snapshot

What generated files look like

A generated migration (toasty/migrations/0000_migration.sql):

CREATE TABLE "users" (
    "id" TEXT NOT NULL,
    "name" TEXT NOT NULL,
    "email" TEXT NOT NULL,
    PRIMARY KEY ("id")
);
-- #[toasty::breakpoint]
CREATE UNIQUE INDEX "index_users_by_id" ON "users" ("id");
-- #[toasty::breakpoint]
CREATE UNIQUE INDEX "index_users_by_email" ON "users" ("email");
-- #[toasty::breakpoint]
CREATE TABLE "todos" (
    "id" TEXT NOT NULL,
    "user_id" TEXT NOT NULL,
    "title" TEXT NOT NULL,
    "completed" BOOLEAN NOT NULL,
    PRIMARY KEY ("id")
);
-- #[toasty::breakpoint]
CREATE INDEX "index_todos_by_user_id" ON "todos" ("user_id");

Each statement is separated by -- #[toasty::breakpoint] markers so the driver can execute them individually within a transaction.

The corresponding snapshot (toasty/snapshots/0000_snapshot.toml) captures the full schema state in TOML:

version = 1

[schema]

[[schema.tables]]
id = 0
name = "users"
primary_key = { columns = [{ table = 0, index = 0 }], ... }

[[schema.tables.columns]]
id = { table = 0, index = 0 }
name = "id"
ty = "String"
storage_ty = "Text"
nullable = false
auto_increment = false

[[schema.tables.columns]]
id = { table = 0, index = 1 }
name = "name"
# ...

A history.toml file ties migrations to their snapshots:

version = 1

[[migrations]]
id = 1101269131896407524
name = "0000_migration.sql"
snapshot_name = "0000_snapshot.toml"

Implementation details

• Schema diffing (toasty-core): Rename-aware diff algorithms for tables, columns, and indices with a RenameHints system to distinguish renames from drop+create
• SQL generation (toasty-sql): New DDL statement types (ALTER TABLE, ALTER COLUMN, ADD/DROP COLUMN, DROP INDEX, COPY TABLE, PRAGMA) with per-database serialization
• Driver capabilities: New SchemaMutations capability struct -- SQLite can't ALTER COLUMN (uses table recreation), MySQL supports atomic multi-property ALTER COLUMN, PostgreSQL requires separate statements per property change
• Driver trait: Extended with generate_migration(), applied_migrations(), and apply_migration(). Migrations tracked in a __toasty_migrations table, applied within transactions. DynamoDB stubbed as unimplemented.
• ~1,880 lines of tests across 6 test files covering create/drop/alter table, add/drop/alter column for SQLite, PostgreSQL, and MySQL

Stats

~6,000 lines added across 70 files (new toasty-cli crate, extensions to toasty-core, toasty-sql, and all SQL drivers)