README.md

jupiter-migrate

SeaORM database migrations for Mega, extracted from jupiter so day-to-day cargo check does not compile migration code unless needed.

Apply migrations

mono enables jupiter/migrate. On startup, Storage::new calls jupiter_migrate::apply_migrations automatically (jupiter/src/storage/init.rs). No separate init CLI step is required.

After schema migrations, mono HTTP boot may run application data backfills (e.g. actor handle → campsite_user_id) using Campsite internal/member_identities and data_backfill_ledger. Deploy campsite-api with MEGA_INTERNAL_SECRET before mono with MEGA_OAUTH__MEGA_INTERNAL_SECRET.

Crates that need a migrated DB in tests should enable jupiter/migrate or ceres feature migrate.

Generate a new migration

cd jupiter-migrate/src/migration
sea-orm-cli migrate generate "your_migration_name"

Commit the new file under jupiter-migrate/src/migration/.

Regenerate entities

After schema changes, regenerate callisto entities (adjust connection URL for your DB):

sea-orm-cli generate entity \
  -u postgres://postgres:postgres@localhost:5432/mono \
  -o jupiter/callisto/src \
  --with-serde both \
  --entity-format dense

Review generated diffs in jupiter/callisto/src/ before committing.

Do not edit CLI-generated entity files for polymorphic/link joins or Model::new helpers — those live only in entity_ext/. Regenerating must overwrite table models cleanly.

After every regen:

1. Re-add pub mod entity_ext; to jupiter/callisto/src/mod.rs (codegen overwrites this file).
2. Keep sea_orm_active_enums.rs webhook variant names readable (ClCreated, not CLI-mangled names). rs_type = "Enum" is correct for SeaORM 2.0 — call sites that need a string use to_value().value or TryFrom<&str>.
3. Leave entity_ext/ alone — it owns:
   • Model::new helpers and ID utilities
   • Polymorphic / link-based Relation + Related (no DB FKs), including:
     • item_labels / item_assignees: dual belongs_to on item_id → mega_cl and mega_issue
     • mega_cl / mega_issue: has_many labels (via), assignees, conversations
     • mega_conversation: link joins to CL/Issue; has_many reactions
     • reactions: belongs_to conversation
     • mega_code_review_thread: belongs_to mega_cl on link
     • label: has_many item_labels

Join call sites that need those relations use callisto::entity_ext::<table>::Relation, not the generated entity Relation.

Library API

use jupiter_migrate::{apply_migrations, Migrator};

apply_migrations(&db, refresh) runs pending migrations. Migrator is the SeaORM migrator trait implementation.