Postgres parameterized interval is passed incorrectly in raw query

Bug description

Consider the following query:

  const endDate = new Date('2020-12-15T06:00:00');
  const startDate = new Date('2020-12-20T06:00:00Z');
  const interval = '1 day';

  const result = await ctx.prisma.$queryRaw`
    SELECT
      start_ts AS timestamp,
      COUNT(*) AS value
    FROM
      generate_series(${startDate}::TIMESTAMP, ${endDate}::TIMESTAMP, ${interval}::INTERVAL) AS start_ts
    LEFT JOIN "Ticket" AS t ON t."createdAt" >= start_ts AND t."createdAt" < start_ts + ${interval}::INTERVAL
    GROUP by start_ts
    ORDER BY start_ts
  `;

This should work, but fails. It only works if the interval is first casted to TEXT and then to INTERVAL:

${interval}::TEXT::INTERVAL

Expected behavior

It should be possible to cast it directly to interval.

Prisma information

@prisma/cli          : 2.12.1
@prisma/client       : 2.12.1
Current platform     : darwin
Query Engine         : query-engine cf0680a1bfe8d5e743dc659cc7f08009f9587d58 (at node_modules/@prisma/engines/query-engine-darwin)
Migration Engine     : migration-engine-cli cf0680a1bfe8d5e743dc659cc7f08009f9587d58 (at node_modules/@prisma/engines/migration-engine-darwin)
Introspection Engine : introspection-core cf0680a1bfe8d5e743dc659cc7f08009f9587d58 (at node_modules/@prisma/engines/introspection-engine-darwin)
Format Binary        : prisma-fmt cf0680a1bfe8d5e743dc659cc7f08009f9587d58 (at node_modules/@prisma/engines/prisma-fmt-darwin)
Studio               : 0.322.0

[emcpow2]

Also getting

Raw query failed. Code: `08P01`. Message: `db error: ERROR: insufficient data left in message

error in this case

        await this.prismaService.$queryRaw`
          SELECT "id"
          FROM "entities"
          WHERE "id" = ${entityId} FOR UPDATE
          `;

the problem might be in entityId: bigint

[elie222]

@JClackett this workaround works from the original message:

where date > current_date - ${period}::TEXT::INTERVAL

Ran into the same issue. Still a problem but double casting not the end of the world. Thanks for the solution @Sytten.