Hi,

I am still working backwards off of a database import but ran into another migration issue.

I built a small test case to reproduce it:

package schema

import (
	"entgo.io/ent"
	"entgo.io/ent/schema/edge"
	"entgo.io/ent/schema/field"
)

// Todo holds the schema definition for the Todo entity.
type Todo struct {
	ent.Schema
}

// Fields of the Todo.
func (Todo) Fields() []ent.Field {
	return []ent.Field{
		field.Int("id"),
		field.String("title"),
		field.Int("author"),
	}
}

// Edges of the Todo.
func (Todo) Edges() []ent.Edge {
	return []ent.Edge{
		edge.From("user", User.Type).
			Ref("todo").
			Required().
			Field("author").
			Unique(),
	}
}

package schema

import (
	"entgo.io/ent"
	"entgo.io/ent/dialect"
	"entgo.io/ent/schema/edge"
	"entgo.io/ent/schema/field"
)

// User holds the schema definition for the User entity.
type User struct {
	ent.Schema
}

// Fields of the User.
func (User) Fields() []ent.Field {
	return []ent.Field{
		field.Int("id").SchemaType(map[string]string{
			dialect.Postgres: "serial",
			dialect.SQLite:   "integer",
		}),
		field.String("name"),
	}
}

// Edges of the User.
func (User) Edges() []ent.Edge {
	return []ent.Edge{
		edge.To("todo", Todo.Type),
	}
}

Follow the tutorial, added a user entity with an edge. Then proceed to add versioned migrations (using atlas to the mix).

So, I execute the migration:

go run -mod=mod ent/migrate/main.go first

And have the following .sql file (I formatted it for easier reading on here):

-- create "users" table
CREATE TABLE "users" (
  "id" serial NOT NULL,
  "name" character varying NOT NULL,
  PRIMARY KEY ("id")
);
-- create "todos" table
CREATE TABLE "todos" (
  "id" bigint NOT NULL GENERATED BY DEFAULT AS IDENTITY,
  "title" character varying NOT NULL,
  "author" serial NOT NULL,
  PRIMARY KEY ("id"),
  CONSTRAINT "todos_users_todo"
    FOREIGN KEY ("author")
    REFERENCES "users" ("id") 
    ON DELETE NO ACTION
);

When I run a migration, I expect:

• two tables (todos, users ✅)
• user.id being a serial (✅)
• todo.author as some kind of integer (🔴)

However, the todos.author is another serial and that seems unnecessary.

On top of that, something I haven't been able to repro with this yet is: in my older software, the initial import shows tables as integer, but with a definition which is equal to my example, it exhibits the same "problem": the first migration will try to convert them to serial and then it wants to add SEQUENCE for each of them.

So imagine this:

1. (import) todo.author = integer
2. first migration (on top of the import):

• alter column integer > serial
• CREATE SEQUENCE IF NOT EXISTS "todos_author_seq" OWNED BY "todos"."author";