Prisma Migrate prevents using special index functionality of CockroachDB.

Problem

We’d like to create an index that uses the STORING clause, available in CockroachDB. Something like this:

CREATE INDEX "Reservation_foo_bar_idx" ON "Reservation"("foo", "bar") STORING ("col1", "col2", "col3", "col4");

We can add an index like this in schema.prisma:

@@index([roomId, endDate])

and then edit the generated migration file to add the STORING clause. But the next time the migrate dev is run, the system will delete the old migration and replace it with one where the STORING clause is not there.

Suggested solution

Extend index configuration to allow a storing field for CRDB, or adjust Prisma Migrate to not delete the index when the introspection doesn’t exactly match on clauses that wouldn’t be recognized by Prisma Migrate anyway.

Alternatives

~Right now, what we’re doing is this:~

// @@index([roomId, endDate])

UPDATE this workaround doesn’t work either, Prisma Migrate will delete it.

[AndrewSouthpaw]

What happens if you do not include the @@index([roomId, endDate]) in your schema - does Migrate still drop it on the next generated Migration @AndrewSouthpaw?

Migrate still drops it at the next generated migration, yeah.

Is it limited to a simple list of “included” columns

AFAICT from the docs and the railroad diagram, that’s correct.

[janpio]

I am closing this here then, as I think #8584 asks for exactly this functionality. What you describe of it dropping it and replacing it with a normal index unless you modify the migration file is expected until that is implemented or we change the behavior otherwise.