Avoiding Hasura sequential scan for query on large array relation - postgresql

Question

How do I get Hasura to generate SQL that uses the index on review.user_id?
Here is the seemingly obvious way of getting a user's reviews. I've included a simplified version of the SQL that Hasura generates, which is a sub query approach rather than a JOIN of the review table on user:
# SELECT * FROM review
# WHERE user_id = (
# SELECT id FROM "user" WHERE username = 'admin'
# )
# LIMIT 50
# This uses a sequential scan on `review` because Postgres
# can't know exactly what the subquery returns.
query ReviewsForUserSlow {
user(where: { username: { _eq: "admin" } }) {
reviews(limit: 50) {
text
}
}
}
Here is a hack to get Hasura to generate SQL that does indeed use the review.user_id index. However, the caveat is that we need to make a round trip to Hasura to get the user ID in order to build this query:
# Simplification of the SQL Hasura generates:
# SELECT * FROM review
# WHERE user_id = '8f3547e4-c8a9-480f-991f-0798c02f2ba2'
# LIMIT 50
query ReviewsForUserFast {
review(
limit: 50,
where: {
user_id: {
_eq: "8f3547e4-c8a9-480f-991f-0798c02f2ba2"
}
}
) {
text
}
}