On conflict on constraint upsert + postgres

Environment

Knex version: 0.95.1 Database + version: PostgreSQL 13.0, compiled by Visual C++ build 1914, 64-bit OS: Win 10

Feature discussion / request

https://knexjs.org/#Builder-onConflict

I haven’t found in the documentation how to upset rows with unique constraints.

Given following constraint:

ALTER TABLE public.assignments_mappings ADD CONSTRAINT assignments_mappings_task_id_user_id_unique UNIQUE (task_id, user_id)

I trying to perform upsert with following code:

 return await db("assignments_mappings")
      .insert(
        tasksID.map(id => ({ task_id: id, user_id, company_id: 1 }))
      )
      .onConflict('assignments_mappings_task_id_user_id_unique')
      .merge()
      .returning('id')

Which generates

error: insert into "assignments_mappings" ("company_id", "task_id", "user_id") values ($1, $2, $3), ($4, $5, $6), ($7, $8, $9), ($10, $11, $12) on conflict ("assignments_mappings_task_id_user_id_unique") do update set "company_id" = excluded."company_id", "task_id" = excluded."task_id", "user_id" = excluded."user_id" returning "id" - column "assignments_mappings_task_id_user_id_unique" does not exist

For successful operation, the output gotta be on conflict on constraint. How can I solve it?

[sjchmiela]

Ok, wrapping lower(email) with db.raw did the trick. Thanks for prompting me to investigate it further!

[abhi-nestegg]

Can anyone confirm that this is the way onConflict will behave in general?

For eg. I am going to use it for an index that is created like so:

CREATE UNIQUE INDEX accounts_unique_account_index ON my_accounts (COALESCE(company_id, user_id), account_id);

So, the corresponding onConflict clause, as per above discussion, will be:

.onConflict([knex.raw("COALESCE(company_id, user_id)"), 'account_id']).merge()

Just want to make sure that it’s a feature and not a side effect.