Migrations suck
🎁Feature Request
I might be off here, but migrations in Convex feel kind of rough.
My workflow: I’m on a dev branch tweaking the schema, I write migrations, and I migrate my local data to test. Then in a later update I need to drop some old fields and make more code changes.
In “classic” SQL, I can bundle schema changes and data migrations together and let the pipeline roll forward automatically. In Convex, it seems like I have to roll the code back to an intermediate state, run the migrations on prod, deploy that intermediate state, and only then make the rest of the code changes again. It works, but it feels pretty clunky.
Am I missing a better pattern for handling schema + data changes without the intermediate deploy step? How are people doing this smoothly in Convex?
My workflow: I’m on a dev branch tweaking the schema, I write migrations, and I migrate my local data to test. Then in a later update I need to drop some old fields and make more code changes.
In “classic” SQL, I can bundle schema changes and data migrations together and let the pipeline roll forward automatically. In Convex, it seems like I have to roll the code back to an intermediate state, run the migrations on prod, deploy that intermediate state, and only then make the rest of the code changes again. It works, but it feels pretty clunky.
Am I missing a better pattern for handling schema + data changes without the intermediate deploy step? How are people doing this smoothly in Convex?
