Migration Question on adding On Delete Cascade

[can-fraction]

Hello,
Thanks for the great package! I'm having some understanding the migration process and would appreciate any help.

I added on delete cascade to my table and want to do migration. I genereted the tests and they failed as expected.

 Schema does not match
workout_log_table:
 columns:
  workout_id:
   Not equal: `NULL REFERENCES workout_table (id) ON DELETE CASCADE` (expected) and `NULL REFERENCES workout_table (id)` (actual) 

When I looked online on how to migrate this, I saw that I have to create a new temptable and switch them however I just randomly added this code and it worked. Tests are passing

      await m.runMigrationSteps(
        from: math.max(16, from),
        to: to,
        // ignore: missing_required_argument
        steps: migrationSteps(
          from16To18: (m, schema) async {
            await m.alterTable(TableMigration(schema.workoutLogTable));
            await m.alterTable(TableMigration(schema.workoutItemTable));
          },
        ),
      );

I'm not sure how my tests passed since I didnt provide any change in the function.
Does this magically migrate or are tests misleading me?
What kind of changes can I make on a table and the alterTable magically migrates?
Sorry if these are basics I read the documentation but couldnt figure out.

[can-fraction]

Oh okay, I got lost in the web documentation but after reading the documentation in the code and reading the code, I realized it does the switch! So I guess my migration actually works. Just quick question can I migrate any table like this since essentially it switches to a new one? Like even if I add a column would the only difference be this function being slower than m.addColumn ?

[simolus3]

Drift's alterTable is kind of a catch-all that will re-create the table, copy data over and then delete the original table. SQLite has builtin support for ALTER TABLE ... ADD COLUMN which will definitely be faster, but is functionally equivalent.

Basically, alterTable works well when changing constraints or when adding new columns where the default data should depend on the existing row (possible with columnTransformer). Changing a column type with a transformer that casts also works with alterTable.

Personally I would always use the most expressive migration to make them easier to understand. So if you're adding a new column, addColumn is IMO a better choice than alterTable. But alterTable is a powerful tool for complex migrations that can't really be expressed without it (and is pretty much necessary when changing column constraints in SQLite).